journal of emerging trends in computing and information sciences · 2011-12-20 · to assist...

Journal of Emerging Trends in Computing and Information Sciences

EDITORIAL ADVISORY BOARD

Prof. Dr. Abderrafiaa Koukam Université De Technologie De Belfort

Montbéliard France

Prof. Amanda Spink Loughborough University

United Kingdom

Dr. Bradford Lee Eden University Of California

USA

Dr. Bernard J. Jansen The Pennsylvania State University University

USA Dr. Bernard T. Han

Western Michigan University USA

Dr. Chia-Nan Wang D&O Biotechnology Co., Ltd

Taiwan

Dr. Clement Leung Hong Kong Baptist University

Dr. David Paper Utah State University

USA

Dr. Donald H. Kraft United States Air Force Academy, Colorado

Springs,Colorado USA

Dr. Hossam Elgindy University Of New South Wales

Australia

Dr.Imtiaz Ahmad Kuwait University

Kuwait

Dr. M. Gordon Hunter The University Of Lethbridge, Alberta,

Canada

Dr. Yacine Lafifi University Of Guelma

Algeria.

Dr. Christos Grecos University Of Central Lancashire.

United Kingdom

Manish Gupta M&T Bank Corporation, Buffalo, New York

USA

Dr. Martin Purvis University Of Otago

New Zealand

Dr. Murali Raman Multimedia University Malaysia,

Malaysia

Dr. Paul Nieuwenhuysen Vrije Universiteit Brussel, Pleinlaan 2, B-1050

Brussel Belgium

Dr. Prasad Bingi Indiana-Purdue University

USA

Dr. Ram B. Misra Montclair State University

USA Dr. Rugayah Gy Hashim

Universiti Teknologi Mara Shah Alam, Selangor Malaysia

Dr. Sajid Hussain Fisk University, Nashville

USA Dr. Satish Kumar Agarwal

University Of Bahrain, Manama Kingdom Of Bahrain

Dr. Eng. Sattar B. Sadkhan University Of Babylon

Iraq Dr. Shaher Momani The University Of Jordan

Amman

Dr. Shakil Akhtar Clayton State University

USA

Dr. Shamkant Madhav Khairnar Maharashtra Academy Of Engineering Alandi

Pin-412105, Pune. Maharashtra State India

Dr. Waleed H. Abdulla The University Of Auckland

New Zealand

Dr. Y. Mustafa Union College

310 College Street Barbourville, Ky

USA

Dr. Yong Seog Kim Utah State University

USA

Prof. Dr. Yong Zhang Shenzhen Universit

China

Yu Zheng Microsoft Research Aisa

Dr. Yucong Duan University Of Bourgogne,

France

Associate Editors

Dr. Dorothea La “Chon” Abraham College Of William & Mary, Williamsburg, Va

USA

Dr. Jyhjong Lin Ming Chuan University

Taiwan

Dr. Luis. C. Rabelo University Of Central Florida - Orlando, Fl

Usa

Dr. Shenping Hu Shanghai Maritime University

China Dr. Shunfu Hu

Southern Illinois University Edwardsville USA

Dr. Sorinel Oprisan College Of Charleston

Rita Liddy Hollings Science Center Charleston, South Carolina

USA

Dr. Vishal Goyal Punjabi University, Patiala

India

Dr. Weifeng Xu Gannon University

USA

Dr. Xiaochun Cheng Middlesex University

United Kingdom

The expertise of editorial board members are called in settling refereed conflicts about acceptance/rejection and their opinion is considered as final.

Table of Contents

Refereed Research Articles Knowledge based Diagnosis of Abdomen Pain using Fuzzy Prolog Rules

Priti Srinivas Sajja, Dipti M Shah 55-60Age Estimation based on Neural Networks using Face Features

Nabil Hewahi, Aya Olwan, Nebal Tubeel, Salha EL-Asar, Zeinab Abu-Sultan 61-68

Impact of Gender and Nationality on Acceptance of a Digital Library: An Empirical Validation of Nationality Based UTAUT Using SEM

Rita Oluchi ORJI 69-79Encouraging the Inclusion of Evolutionary Psychology into Information Systems’ Theoretical Repertoire as an Emerging Trend

Chon Abraham 80-89Fruit Recognition using Color and Texture Features

S.Arivazhagan, R.Newlin Shebiah, S.Selva Nidhyanandhan, L.Ganesan 90-94Solving Fuzzy Based Job Shop Scheduling Problems using Ga and Aco

Surekha P, S.Sumathi 95-102A Systems Approach for Dealing with Resistance to Change: With Reference to Library and Information Professionals Working in Academic and Research Sector Libraries in India

Kshema Prakash 103-116The Rule Based Intrusion Detection and Prevention Model for Biometric System

Maithili Arjunwadkar, R.V. Kulkarni 117-120Using Multimodal Fusion in Accessing Web Services

Atef Zaguia, Manolo Dulva Hina, Chakib Tadj, Amar Ramdane-Cherif 121-137

VOL. 1, NO. 2, Oct 2010 E-ISSN 2218-6301 Journal of Emerging Trends in Computing and Information Sciences

©2009-2010 CIS Journal. All rights reserved.

http://www.cisjournal.org

Knowledge based Diagnosis of Abdomen Pain using Fuzzy Prolog Rules

Priti Srinivas Sajja, Dipti M Shah

Department of Computer Science & Technology Sardar Patel University, Vallabh Vidyanagar

Gujarat, India Email: [email protected]

ABSTRACT

Decision support through Information Technology is an integral part of our lives. It is being increasingly used for decision-making in the medical science also. This article introduces the information explosion in medical field elaborating the need of a knowledge-oriented decision support system for diagnosis of abdomen pain. Main objective of the system is to assist doctors, assistants and social workers in their decision making process and create awareness in the area especially where trained manpower is in scarce. To impart the fuzziness of the domain, modified Prolog rule format is used, which is illustrated in a case of appendicitis. This article presents general framework of system, sample rules, resulting charts and sample screens of the prototype implementation. Keywords: Knowledge Based Diagnosis, Advisory System, Fuzzy Prolog, and Decision Tree. 1. INTRODUCTION

There is an increasing appreciation of the role

that computers and informatics are playing to improve the overall health delivery systems. Diagnosis requires far more critical decision-making on a wide range of options. It also requires a large amount of humanism [1]. Medical informatics has high visibility through applications in areas such as diagnostic techniques in ultrasonography, x-ray, computerized tomography scanning, nuclear magnetic resonance imaging etc. The other areas can be clinical laboratories, pathological investigations and computer-assisted decision-making. In medical practice, data acquisition as well as subsequent storage, retrieval and manipulation of the data are enhanced by efficient computerization through database in static fashion. Such decision-making through computers is cost effective and improves accuracy [2]. However, the decision support should be knowledge-oriented to improve effectiveness of the decisions made. Knowledge-oriented decision-making by its nature helps in identifying most plausible diagnosis and provides ease of choosing an appropriate treatment. Large amount of existing medical knowledge and rapid growth of the knowledge have resulted in a situation where even specialists find it increasingly difficult to assimilate and use the information that would be useful in making effective decisions. This leads the decision making a tedious and time-consuming process.

Most of the Indian population lives in rural areas where doctors and specialists are in scare. Due to scarcity of experts and information explosion in the field of medicine, there is a need of knowledge-oriented decision-making systems. The main objective of the system is to assist doctors, assistants and social workers in decision-making process for various kinds of the abdomen diseases. Information on the cause, diagnosis, symptoms, complications, prevention and treatment needs to be identified, documented and inferred by the system for benefits of the rural locality; especially where expertise is

not available on demand. This type of system will be an efficient means to store and to pass experts knowledge in documental form for long time and it can provide primary advice to the health workers and patients in initial stage. Hence, an advisory system that applies the knowledge in diagnosis and determines patient’s condition is the prime necessity in the medical field. This article proposes a general framework of knowledge-oriented decision support system for advisory, diagnosis and awareness for the field. As an illustration, a few sample fuzzy rules are formed with the help of domain experts for appendicitis case. 2. KNOWLEDGE-ORIENTED APPROACH

A Knowledge-Based System (KBS) has

interdisciplinary approach of various disciplines like computer science, cognitive science, hardware field etc. The society and industry are becoming knowledge-oriented and rely on different experts’ decision-making abilities depending on the information available. When expertise is unavailable, a KBS can act as an expert on demand to save time.

KBS can save money by leveraging expert, allowing users to function at higher level and promoting consistency [3]. One may consider the KBS as a productive tool, having knowledge of more than one expert for long period of time. Large percentage of the population in India lives in rural and remote areas, where medical facilities are unavailable. Hence, there is a need of a system that supports decisions and increases awareness in the area. Such system can potentially alleviate the immense diagnostic workload of rural health workers and medical practitioners. Also, it can assist in prevention of various diseases. In addition, by using such systems, user will get the advantage of the knowledge of more than one specialist. Such KBS uses Artificial Intelligence techniques for efficient and effective decision making in unstructured domain and apply reasoning and

55

mailto:[email protected]




explanation facility for the domain problem to achieve high level of performance.

The large amount of existing medical knowledge and the rapid growth of that knowledge during the last quarter of the 20th century resulted in a situation where most clinicians find it increasingly difficult to assimilate the field information, which could be useful in making optimal clinical judgment. The system can provide a solution to much of the problems created by such information explosion. Decision making by the clinician in the management of patient’s data is a highly intellectual activity, which involves: (i) skill in gathering and evaluating the information about the patient; and (ii) ability to effectively utilize the large body of medical

knowledge. The system can facilitate to improve the clinician’s performance of each of this task. 3. STRUCTURE OF THE SYSTEM

The model of the discussed system is given in

Figure 1 representing the overall process structure of KBS for medical diagnosis. The basic components of the system are the knowledge base, inference engine, and a workspace. The knowledge base of the system plays a key role in the procedure of decision-making [4] by efficiently storing the domain knowledge and patients history. Temporary results can be stored in workspace. The inference engine is a program, which infers the knowledge available in the knowledge base [5].

The knowledge base of the proposed system first

receives the preliminary information from the patient through self-administrated questionnaires and retrieves the patient’s history, if any, stored in the knowledge base. After evaluating the inputs, the program presents a new questionnaire to the patient. Using the information, the system suggests action/advisory for further tests and/or conclusion about the patient’s disease. An expert may have a manual control/justification of the given suggestions and alternatives depending on degree of uncertainty associated with the patient’s response and overall decision-making process.

The knowledge of the expert in the decision-making can be represented in various forms. The knowledge of expert can be easily represented into rule-based format as a set of conditional rules [6]. Rules may be chained according to the knowledge it represents. Considering the uncertainty of the diagnosing process, the fuzzy rules are used here. To accommodate such fuzziness, typical Prolog rule format is modified. Each rule has a basic form -

IF antecedents THEN consequent

If certain antecedents are evaluated as True, then it logically follows the consequent. As denoted above, the modified Prolog rule format is Hypothesis (Name, Disease, Probability); Symptom (Name, Indication, Probability); Symptom (Name, Indication, Probability); … Symptom (Name, Indication, Probability). .................................................... Eq. (1)

Here Hypothesis and Symptoms are user defined predicates in Prolog [7]. These predicates use symbols (variables) like Name (user name), Disease (disease name) and Probability (chances in percentage). Probability factors given along with the rules for the concerned advises are considered as the degrees of uncertainty related with the decision taken [8]. These values are determined by taking samples from experts. The hypothesis proved true if patient’s data has all the indications given in symptom lists. Systems knowledge base consists of such multiple fuzzy rules representing the domain knowledge in the form of Prolog code.

56




4. AN ILLUSTRATIVE CASE OF AN

APPENDICITIS PROBLEM

An intelligent advisory system for abdomen pain can also take care for the appendicitis problem including various other abdomen diseases. Interactive sample rules are proposed here, that can directly assist doctors or other health workers in finding the probability of having particular disease like appendicitis.

Appendicitis usually commences as in inflammation of the mucous membrane or lymph follicles, which may terminate in one of the following ways [9]:

Resolution Ulceration Suppuration Gangrene Fibrosis

The system prepares the history of patient by asking several questions in the language understood by layman and predicts the probability of having a disease and presents advice for further treatment. 4.1 Sample Rules for the Above Case

Sample rules in modified prolog for the prototype implementation of the system are given as follows: R1 Hypothesis(Patient, appendicitis, 0.1) :- symptom(Patient,pain_ur, 0.4), symptom(Patient, pain_shift_rif, 0.1), symptom(Patient, o_test_result, 0.2). R2 Hypothesis(Patient,rs_test, 1.0) :- symptom(Patient,pain_rif, 1.0), symptom(Patient, vomit, 0.8), symptom(Patient, less_temp, 1.0),

symptom(Patient, tenderness_site, 0.8). R3 Hypothesis(Patient, appendicitis, 0.6) :- symptom(Patient,pain_ur, 0.7), symptom(Patient, o_test_result, 1.0). R4 Hypothesis(Patient,o_test, 1.0) :- symptom(Patient,pain_rif, 1.0), symptom(Patient, vomit, 0.8), symptom(Patient, less_temp, 1.0), symptom(Patient, tenderness_site, 0.8). R5 Hypothesis(Patient, appendicitis, 0.4) :- symptom(Patient,pain_ur, 0.7),

symptom(Patient, o_test_result, 0.3). R6 Hypothesis(Patient,p_test, 0.8) :- symptom(Patient,pain_ur, 0.7), symptom(Patient, pain_shift_rif, 0.6), symptom(Patient, vomit, 0.6),

symptom(Patient, less_temp, 0.8), symptom(Patient, tenderness_site, 0.5). R7 Hypothesis(Patient, appendicitis, 1.0) :- symptom(Patient,pain_rif, 1.0),

symptom(Patient, o_test_result, 1.0). R8 Hypothesis(Patient,p_test, 1.0) :- symptom(Patient,pain_ur, 0.7), symptom(Patient, pain_shift_rif, 0.6), symptom(Patient, vomit, 0.8), symptom(Patient, less_temp, 0.8),

symptom(Patient, tenderness_site, 0.6). R9 Hypothesis(Patient, appendicitis, 1.0) :- symptom(Patient,pain_rif, 1.0),

symptom(Patient, rs_test_result, 1.0). R10 Hypothesis(Patient,p_test, 1.0) :- symptom(Patient,pain_rif, 1.0), symptom(Patient, vomit, 0.8), symptom(Patient, less_temp, 1.0), symptom(Patient, tenderness_site, 0.8). R11 Hypothesis(Patient, appendicitis, 1.0) :- symptom(Patient,pain_rif, 1.0),

symptom(Patient, p_test_result, 1.0). R12 Hypothesis(Patient, appendicitis, 0.1) :- symptom(Patient,pain_ur, 0.2), symptom(Patient,pain_er, 0.2), symptom(Patient, pain_rif, 0.1), symptom(Patient, vomit, 0.8), symptom(Patient, less_temp, 0.8), symptom(Patient, o_test_result, 0.2).

As stated above, the knowledge collected from the field expert is codified in the Prolog language to form a rule base for the application. These rules are sequentially executed to come to a concluding/diagnosis. If the mentioned symptoms match, the hypothesis of having the appendicitis is true to some extent. For example, according to rule 6 (R6) ‘If patient has pain in Umbilical Region and the pain shifts to Right Illias Fossa Region along with vomiting and tenderness symptoms at the site of appendicitis’ (Region 4 in Figure 3), then Psoas Test is recommended with 80% probability. When rule 6 (R6) is true, the system will check all rules sequentially and fires rule 11 (R11) concerning the Psoas Test result in the symptom and conclude the probability of appendicitis.

Such multiple rules can be developed from a decision tree by maintaining the decision sequence. This is illustrated in Figure 2 in the form of decision tree that has critical information with proper directions for the discussed case. For structured decision-making, the whole abdomen region is divided into nine major parts as shown in Figure 3. For each of these 9 parts a separate tree can be prepared. Figure 4 shows results of the prototype implementation using the above discussed rules.

57




58




Figure 4. Resulting output of the experiment

5. CONCLUSION

The paper discusses architecture, and an

experiment of abdomen pain using a model of a KBS for medical diagnosis. In similar manner, such system can be enhanced further for all possible cases in abdomen pain. Different types of customized heuristic can be incorporated with the fuzzy prolog rule structure used here.

This type of system can be proved an efficient means to store and pass experts knowledge in documental form for long time. Hence it can be used as a training and documentation tool also.

Using the proposed design one may go for the diagnosing and advisory systems in different domain. Development of an editor which enables knowledge engineer or expert to edit knowledge within the framework leads to a generic commercial product for KBS in diagnosing and advisory category.

ACKNOWLEDGMENT The authors are thankful to Dr. Balmukund Shah,

Surgeon, Krishna Hospital, Borsad, India for providing the

domain knowledge and necessary support for automation of the decision making process of the system. REFERENCES [1] R.D. Lele, Computer in Medicine. Tata McGraw

Hill Publishing Company Ltd., New Delhi, 1988.

[2] D.M. Shah & S.M. Patel, IT in Healthcare. Presented at the Proceeding of the National Seminar on Computer Applications, (2001) Vallabh Vidyanagar, India.

[3] P. Harmon, R. Maus & W. Morrissey, Expert Systems Tools & Applications. John Wiley & Sons, NY, USA, 1988.

[4] E. Turban, Decision Support and Expert Systems. Mac. Publication Company, 1993.

[5] D.A. Waterman, A Guide to Expert Systems. IEEE Press, NY, USA, 1988.

[6] L. Brownston, R. Farrell, E. Kant & N. Martin, Programming Expert Systems in OPS5, An Introduction to Rule Based programming. Addison Wesley Publishers, USA, 1986.

59




[7] W.F. Clockin & C.S. Melish, Programming in Prolog. 5th Edition, Springer-Verlag, NY, USA, 2003.

[8] D.M. Shah & P.S. Sajja, A Knowledge-Based Decision Support System for Diagnosis of Abdomen Pain. Prajna, 15, (2007), 28-35.

[9] T.N. Patel, A System of Surgical Diagnosis. BI Publications Ltd, Mumbai, 1985.

60




Age Estimation based on Neural Networks using Face Features

Nabil Hewahi, Aya Olwan, Nebal Tubeel, Salha EL-Asar, Zeinab Abu-Sultan

Corresponding Author: Faculty of Information Technology Islamic University of Gaza - Palestine

ps.edu.nhewahi@iugaza: Email

ABSTRACT In this paper we propose a methodology based on Neural networks to estimate human ages using face features. Due to the difficulty of estimating the exact age, we developed our system to estimate the age to be within certain ranges. In the first stage, the age is classified into four categories which distinguish the person oldness in terms of age. The four categories are child, young, youth and old. In the second stage of the process we classify each age category into two more specific ranges. The uniqueness about our research project is that most of the previous research work do not consider the fine tuning of age as we are presenting in our research. Our proposed approach has been developed, tested and trained using the EasyNN tool. Two public data sets were used to test the system, these are FG-NET and MORPH. To evaluate our system’s performance, we carried out a comparative study between our proposed system, human being and other research trails. The obtained results were significant.

Keywords: Age estimation, Neural networks, face feature extraction.

1. INTRODUCTION Recognition of the most facial variations, such as

identity, expression and gender, have been extensively studied. Automatic age estimation has rarely been explored. In contrast to other face variations, aging variations presents several unique characteristics which make age estimation a challenging task. Since human faces provide a lot of information, many topics have drawn attention and thus have been studied intensively. The most prominent thing of these is face recognition. Other research topics include predicting feature faces, classifying gender, and expressions from facial images, and so on. However, very few studies have been done on age classification or age estimation. In this research, we try to prove that computer can estimate/classify human age according to features extracted from human facial image using Artificial Neural Network(ANN or NN).

Facial features were used in many researches such as recognition, classifying gender, expressions and so on. But few of them have been done on age classification especially on age estimation. Many attempts towards age estimation are tried and most of them give results for wide ranges of ages, or classify the ages in categories such as child, young, youth and old[1][2][3][4]. The problem of having an appropriate approach for age estimation for getting more specific categories of age ranges is still a challenging problem. Thus, we focus our research on more specific age ranges. Actually this problem is also a human problem where many people miss estimate the human ages. To achieve our research goal, we have to find a good database that we can use to test and train our proposed approach, also we have to construct a proper ANN to model our problem. Developing such kind of systems might help in many security purposes or in cases of having disabled people (dump and deaf people). Because our main goal is age estimation and not face recognition, we care only about images of front image, with face free from classes or beard. 1.1 AGE ESTIMATION

Age estimation is the determination of a person's age based on biometric features. The determination of the

age of a person from a digital photography is an intriguing problem, It involves understanding of the human aging process. People cannot freely control aging variation, the collection of sufficient data for age estimation is extremely laborious. Estimating of exact age is one of the most difficult problems even for human being. Therefore, most of the researcher who are working on age estimation are trying to get the results in certain age ranges. The experimented age ranges are still considered to be wide and in some cases exceed 10 years while in other cases reach 15 or 20 years. One of the main problems to reduce the size of the age ranges is how correct and comprehensive the extracted features from the face are. Some researchers use 20, 22, 35,or 68 features, and the accuracy of the results vary depending on the extracted features and the used approach for age estimation. There are some open databases used for testing age estimation systems such as FG-NET [5][6[7] and Morph [8].These datasets contain photos and ages of the people and there are usually ages from 1 year to 70 years. 2. RELATED WORK

Human age estimation by face images is an interesting yet challenging research topic emerged in recent years. There are some earlier works which aimed at simulating the aging effects on human faces, which is the inverse procedure of age estimation.

Horng, et.al [4] proposed an approach for classification of age groups based on facial features. The process of the system was mainly composed of three phases: Location, feature extraction and age classification. Two backpropagation neural networks were constructed. The first one employs the geometric features to distinguish whether a facial image is a baby or not. If it is not, then the second network uses the wrinkles features to classify the image into one of three adult groups. This approach is somewhat efficient as it has 99.1% verification rate for the first network and 78.49% for the second. Fukai, et.al [1] proposed an age estimation system on the AIBO (AIBO is autonomic entertainment robot produced by SONY). HOIP face image database and images captured using

61











AIBO were used, they used Genetic Algorithms (GA) as optimization approach to select important features, then they used Self Organizing Map (SOM) algorithm in training the neural network. Age estimation error of this approach is 7.46% which is considered to be reasonable. Geng, et.al [2] proposed a subspace approach named AGES (Aging pattern Subspace) for automatic age estimation. The basic idea of AGES is to model the aging pattern, which is defined as a sequence of a particular individual’s face images sorted in time order, by constructing a representative linear subspace. The proper aging pattern for a previously unseen face image is determined by the projection in the subspace that can reconstruct the face image with minimum reconstruction error, while the position of the face image in that aging pattern will then indicate its age. The Main Absolute Error of this approach is 6.77% which is better than age estimation system of the AIBO and the ages in AIBO system are between 15-64, but in AGES they are between 0-69. Geng, et.al [3] extended their previous work on facial age estimation (a linear method named AGES). In order to match the nonlinear nature of the human aging progress, a new algorithm named KAGES is proposed based on a nonlinear subspace trained on the aging patterns, which are defined as sequences of individual face images sorted in time order. In the experimental results, the Main Absolute Error is 6.18% that is better than all the compared algorithms but with minor difference with AGES approach. Luu, et.al [8] introduced a novel age estimation technique that combines Active Appearance Models (AAMs) and Support Vector Machines (SVMs), to dramatically improve the accuracy of age estimation over the current state-of-the-art techniques. Characteristics of the facial input images are interpreted as feature vectors by AAMs, which are used to discriminate between childhood and adulthood, prior to age estimation. Age ranges are between 0-69.The Main Absolute Error is 4.37 % that is better than all previous works. It is worth mentioning that in all the used approaches, the age range classifications are not as we are planning to consider in our classifications. We aim at tunnnig the age ranges in more specific domains as we will explain in the next sections.

3. THE PROPOSED APPROACH The proposed system is mainly using a supervised neural networks with backpropagation algorithm. the image is entered to the system, features are extracted, the image is classified in one of the four main age classes, then a more specific age range class is specified. This process is shown in Figure 1.

The main process of our proposed system is shown in Figure 2. We firstly classify age into four main age categories and each age category is classified into two age ranges. We obtained our data to train and test our system from two databases, FG-NET [7] and MORPH [9]. The images in FG-NET are ready and their features are already extracted. MORPH database have only images with some other related information, but without extracted features. For this purpose we ought to extract the features from the images obtained from MORPH database using am-markup tool. Finally we train the system with Easy-NN tool based on the two datasets FG-NET and MORPH.

3.2 AGE CLASSIFICATION PHASE As mentioned before, most of the researchers

categorize the ages into four classes, childhood, young, youth and old. This classification is more or less as our classification in the main classification stage shown in Figure 3. In our system we go further by classifying each of the main categories (each class) into two classes which we call secondary classification stage. The classification in the secondary stage are not partitioned equally, instead the age partitions are based on some changes in the facial features. For example the face features for people who are between 13 to 25 are very close. Also the face features of the ages from 36 to 45 are almost negligible. In some cases we make the range a little bit wide because in many cases it is really difficult to categorize the age in a smaller range. 3.3 DATA COLLECTION PHASE Getting direct photos and collecting the features from each photo are tiresome. To train and test our system, we used ready datasets organized in FG-NET and MORPH. The FG-NET contains 68 features, where each is a pair of points. The collection of features represent the mouth, nose, eye and the face surroundings. Because the number of photos in FG-NET is not enough for our testing and training, we used also MORPH data set. The problem with the MORPH data set is that the features are not extracted. Therefore, we first extracted the features of the photos according to features in FG-NET and put both the data sets in one file. The used points landmarks are shown in Figure 4[10]. Table 1 illustrates each point ‘s location on the face. We have chosen 500 images from the FG-NET. The image should be in front face ,clear and free from glass ,beard and moustaches. Only those images are chosen because we are concerned with age estimation and not face recognition. We felt that 500 images are not enough for training and testing, and they do not cover all the ages, therefore, we were forced to use also another database, namely MORPH. MORPH data corpus comprises facial images of numerous individuals and includes essential metadata, such as age, sex, ancestry, height, and weight. We have chosen 131 image from MORPH [9]. To locate the face features on the images obtained from MORPH dataset, we used am_markup tool [11] which is considered to be the appropriate tool to locate the necessary points on the face images and extract the features. We then combined the two files into one file to be used for training and testing. After collecting all the data in one dataset file (main set), we constitutes four other data sets out of the main set. Each constituted set is used for training and testing one of the secondary stage of ages. The number of examples in each of the used data sets is shown in Table 2.

62




Tab

Figure 3. Age classification based on the proposed model.

Face

Face Bott Top Lip Nos Left Righ Left Righeyeb Left Righ Nos

Image is entered

Feature Extraction

First stage classification using NN

Second stage classification using NN

Figure 1. The stages of the proposed system.
Figure 2.The process of the proposed model
le 1. Facial landmarks

regions Point Id Number of points

edges 0-14 15 om lip 55-59 5 lip 49-53 5 Outline 48,54,60-66 9 e 37-45,67 10 eye 27-30 4 t eye 32-35 4

eyebrow 21-26 6 t row

15-20 6

iris 31 1 t iris 36 1

trils 46,47 2

Figure 4. Points as landmarks according to FG-NET[10]

Table 2. Each used dataset with its corresponding number of examples

Number of examples Data Base

631 examples Main DB

261 examples First

Secondary DB156 examples Second

Secondary DB138 examples Third

Secondary DB76 examples Fourth

Secondary DB

63




A sample of the data features in the main data base are shown in Table 3.

Table 3. Sample of data features.

3.4 NEURAL NETWORK STRUCTURE

To achieve our target, we constructed five NNs. The first NN is the main neural network. The main NN is concerned with the main age classification with four outputs as explained before. The other four NNs are concerned with secondary age classification. Each neural network (the five networks) has 68 pairs of inputs representing the face features in addition to the gender of the person. It is also to be noted that we have two hidden layers, one hidden layer is to correlate each pair in one meaningful unit and the second is considered to be the real hidden layer after organizing the input data in the first hidden layer. The gender is connected directly to the second hidden layer. The number of outputs for each NN related to secondary classification is two representing the more specific age range in its domain. Table 4 depicts the number of outputs and ages categories in each of the five NNs.

Table 4. The number of outputs for each NN and age

classification

Neural Networks Outputs Neural Network

Four output (1-12 , 13-25, 26-45 , 46-63)

Main NN

Two outputs( 1-8 , 9-12 ) 1ST Secondary NN

Two outputs(13-19,20-25) 2nd Secondary NN

Two outpus( 26-35 , 36-45 ) 3rd Secondary NN

Two outpust(46-53 ,54-63 ) 4th Secondary NN

Figure 5 shows the structure of the main NN, whereas Figure 6 shows the structure of 1st secondary NN as example for other secondary NNs.

To implement our system , we used Easy-NN as a tool to build , develop, train and test our framework. Each data set is partitioned into two parts, a part for training and another for testing. In general about 2/3 of the data is used for training where the rest is used for testing.

4. EXPERIMENTAL RESULTS

Ages in both databases are distributed in wide ranges: 1-63 for FG-NET and 26-63 for MORPH. The face features used in the experiments are Sixty-eight landmark points of each face image as presented and shown before.

Many experiments have been tried to get acceptable results in the main and secondary age classification stages, the main stage contains four ranges

and each secondary stage contains two ranges. Many experimentations have been done to get the best results. This is performed by trying various NN parameters such as the momentum that can be changed to any value from 0 to 0.9, Learning Rate that can be changed to any value from 0.1 to 1.0 , and the number of hidden layer nodes. 46-

63 26- 45

13- 25

1- 12

Gen-der

a136... …a2 a1

F F F T M 216.782 200.858 74.933

F F T F F 302.419 251.5 61.096

Figure 5. The structure of the main NN used for the main

four classifications.

Figure 6. The structure of the 1st secondary N

Figure 7. Learning progress for the main NN

1-8

9-12

N.

64




4.1 EXPERIMNTAL RESULTS FOR THE MAIN NN IN THE MAIN STAGE In this experiment we have trained the main NN with specific parameters. Age range in this stage is (1-63) .The values of the used parameters and the corresponding results are shown in Table 5.

Table 5. Details for the main NN parameters and results

The learning progress graph for the main NN is shown in Figure 7. The figure shows the maximum, average and the minimum training error. The average validating error is shown if any validating examples rows are included. From Figure 7, it is noticed that the minimum validating error is in cycle 197. For sure, if the results were taken in that stage, the main NN would have been performed much better than what we obtained. Moreover, it is also expected to get better validating error if we do more cycles (more than 775).

4.2 EXPERIMENTAL RESULTS FOR THE 1ST NN IN THE SECONDARY

In this experiment we have trained the first secondary stage with specific parameters, age range in this stage is (1-12). Table 6 shows the used parameters and the corresponding results.

4.3 EXPERIMENTAL RESULTS FOR THE 2ND NN IN THE SECONDARY STAGE In this experiment we have trained The second NN in the

secondary stage with specific parameters, age in this stage is (13-25) . Table 7 shows details for the used parameters and the corresponding results. In this experiment the result was not so good (about 77%) because the face features were very similar in the age range, and it is difficult to recognize the human age in this stage.

Table 6. Details for the 1st NN in the secondary stage

Detailed information for the 1st NN in the secondary stage

Learning Rate

0.3

Momentum

0.3

Training Error

0.051573

Validating Error

0.063607

Target Error

0.01

Training Example

171

Validating Example

90

Validating Result

92.86%

2nd Hidden Layer Node

28

Cycle

5100

Detailed information for the main NN

Learning Rate

0.3

Momentum

0.3

Training Error

0.033980

Validating Error

0.054266

Target Error

0.01

Training Example

421

Validating Example

210

Validating Result

85.24%

2nd Hidden Layer Node

36

Cycle (Epoch)

775

Table 7. Details for the 2ed NN in the secondary stage.

Detailed information For the 2ed NN in the secondary stage

Learning Rate

0.2

Momentum

0.2

Training Error

0.09

Validating Error

0.20

Target Error

0.01

Training Example

101

Validating Example

55

Validating Result

76.92%

2nd Hidden Layer Nodes

28

Cycle

6874

4.4 EXPERIMENTAL RESULTS FOR THE 3RD NN IN THE SECONDARY STAGE In this experiment we have trained the third secondary stage with specific parameters, age in this range is (26-45). Table 8. shows the used parameters and the corresponding results.

65




Table 8. Details for the 3rd NN in the secondary stage

Detailed information for the 3rd

NN in the secondary stage Learning Rate

0.3

Momentum

0.3

Training Error

0.09

Validating Error

0.10

Target Error

0.01

Training Example

92

Validating Example

46

Validating Result

89.13%


29

Cycle

3000

4.5 EXPERIMENTAL RESULTS FOR THE 4TH NN IN THE SECONDARY STAGE

In this experiment we have trained the third secondary stage with specific parameters, age in this range is (46-63).Table 9 shows the used parameters and the corresponding results.

Table 9. Details for the 4th NN in the secondary stage

Detailed information for 4th NN in

secondary stage Learning Rate

0.3

Momentum

0.3

Training Error

0.1994

Validating Error

0.1848

Target Error

0.01

Training Example

54

Validating Example

22

Validating Result

86.36%


26

Cycle

12456

5. COMPARATIVE STUDY This section discusses comparative study between

our proposed system, human observation and previous work.

5.1 HUMAN OBSERVATIONS

Since one of our research targets is to test our system and compare it with human being, we took opinions of fourteen people for 78 images selected from all ranges randomly. It is interesting to see how much far or close our system to human being is. Based on human being decisions, the correct rate for the first range (1-12), second range (13-25), third range (26-45) and fourth range (46-63) are 87.6%, 67%, 86.4% and 70.9% respectively. It is to be noted that the more specific age ranges (second stage classification) within each of the first stage age classification is considered. The overall correct rate for the human being is 77.9% which is lower than our rate (i.e, 82.3%). It is to be noted that people have the most confusion in the second and fourth ranges whereas our system is confused mostly in the second age range. It is generally clear that our proposed system outperforms the human being. 5.2 OTHER SYSTEMS Because none of the previous research work is focused on more tuned age ranges as we have done, the comparative study becomes a little bit difficult. Comparing our results with the work presented in [6 ] where the authors first classify the input to baby or others and then try to classify others to three age categories. In the first stage the accuracy is very high (up to 99%), and this is simple to obtain because the classification is only to recognize whether the given input is for a baby or others. In the second stage, the new classifier tries to classify the others in one of three age ranges and in this case the performance does not exceed 78.4%. Thus the overall performance is about 88%, but the error rate for the others is very high and if more ages division are considered as we have done, the error rate would be absolutely more. Moreover, the baby age range is just in the range of 1-2 years. This means the 88% does not really scale the overall system performance. If in our case results, we want to consider the others results neglecting the babies, the system performance would be 84% with more correct age ranges which is higher than 78.4% in [6]. This demonstrates that our proposed system can better recognize the others than what is proposed in [ 6].

6. CONCLUSION We proposed an approach for age estimation using facial features based on neural networks. We classified the ages firstly into four categories, then each age range category is also classified into two more specific age ranges. This had not been done before elsewhere. We used five neural networks to achieve our task. The facial features rely on 68 landmark points taken from face images. The development process includes age classfication, data collection ,feature extraction by markup tool and finally training and testing the system by EasyNN. To train and test our system, we used two datasets, FG-

66




NET and MORPH. A comparative study has been conducted between the performance of our proposed system and human being. The proposed system outperforms the human being where the human being performance is 78% whereas the proposed system performance is 82.3%. Also the proposed system is compared with the work proposed in [6] and again the proposed system outperforms the other system where the performance of the system in [6] for others (age categories except the baby age range) is 78.4% whereas in our system, it is 84%. The main future directions are 1. Joining the work of the two stages neural network classifiers. 2. Capturing a real human face image and estimate its age using our proposed system 3. Optimizing the number of face landmark points.

REFERENCES [1] Fukai, H., Nishie, Y., Abiko , K., Mitsukura, Y.,

Fukumi, M. and Tanaka, M. "An Age Estimation System on the AIBO", International Conference On Control, Automation And Systems, pp.2551-2554, 2008.

[2] Geng, X., Zhou, Z. and Smith-Miles, K. "Automatic

Age Estimation Based on Facial Aging Patterns", IEEE Transaction On Pattern Analysis And Machine Intelligence, Vol. 29, No. 12, pp.2234-2240, December 2007.

[3] Geng, X., Smith-Miles, K. and Zhou, Z. "Facial Age

Estimation by Nonlinear Aging Pattern Subspace", Proceedings Of The 16th ACM International Conference on Multimedia , pp. 721-724, 2008.

[4] Horng, W., Lee, C. and Chen, C. "Classification of Age

Groups Based on Facial Features", Journal of Science and Engineering, Vol. 4, No. 3, pp. 183-192, 2001.

[5] Cootes , F., C. J. Taylor, D. H. Copper, and J. Graham,

"Active Shape Models – Their training and Application", Computer Vision Graphics And Image Understanding, Vol. 61, No 1, pp. 38–59, 1995.

[6] Cootes ,F. , Gareth J. Edwards , Christopher J. Taylor,

"Active Appearance Models", Proceedings of the 5th European Conference on Computer Vision-Volume II, p.484-498, June 02-06, 1998.

[7] http://www.fgnet.rsunit.com . [8] Luu, K., Ricanek Jr., K. and Y. Suen, C. "Age

Estimation using Active Appearance Models and Support Vector Machine Regression", Proceedings Of The IEEE International Conference On Biometrics :Theory Applications And System , pp.314-318 ,2009.

[9] Ricanek Jr., K and Tesafaye, T., “ MORPH: A

Longitudinal Image Database of Normal Adult Age-Progression", Proceedings of the 7th International Conference on Automatic Face and Gesture Recognition, p.341-345, April 10-12, 2006.

[10]http://personalpages.manchester.ac.uk/staff/timothy.

f.cootes/data/xm2vts/xm2vts_markup.html 9 [11]http://www.scholarpedia.org/article/Facial_Age_Estimation

67

http://personalpages.manchester.ac.uk/staff/timothy




Impact of Gender and Nationality on Acceptance of a Digital Library: An Empirical Validation of Nationality Based UTAUT Using SEM

Rita Oluchi ORJI

Computer Science Department University of Saskatchewan, Canada

E-mail: [email protected]

ABSTRACT

Electronic Library Systems (ELS) has become an inevitable part of educational institutions. Though millions of dollars are spent building and developing these systems, research findings indicates that millions of potential users may still be ignoring them. Consequently, different Technology Acceptance Models have been applied towards understanding the effect of various factors on end users acceptance of Information Systems (IS). Gender has been found to be an important factor and, as such, has attracted a lot of attention from the research community. Our research, however, recognized the existence of a different cohort group of users of ELS with different usage behavior not just based on gender but also based on nationality and, therefore, developed a Nationality Based UTAUT (NUTAUT) adapted from UTAUT to account for the effect of gender and nationality on acceptance, simultaneously. Nationality was introduced based on the assumption that the UTAUT independent constructs will impact on acceptance and usage differently when moderated by gender and nationality simultaneously. The result obtained from the analysis of data collected from 116 student participants provides support for NUTAUT by showing that the various UTAUT constructs exert varying degree of effects when moderated by gender and nationality combined. It not only confirms the NUTAUT robustness in predicting acceptance of both Male and Female Students (91% and 85% respectively) but also determines the importance of each independent construct to each group, based on their nationality and gender. Social Influence was found to be significant for both international males and females while effort expectancy is a more significant factor for both national males and females. This result indicated that the effect of gender on adoption and use of technology also differ based on nationality. The results of the study are expected to assist in understanding the use of ELS across different cohort groups in society, particularly those in different gender and nationality groups, and serve as a mechanism in guiding the development of ELS, as well as in aiding policy formulation regarding ELS and IS design for schools, research, commercial and work places. Structural Equation Modeling (SEM) was used as the main technique for data analysis.

Keywords: Digital Library, Technology Acceptance, Information Retrieval, UTAUT Model, Structural Equation Model, Gender, Nationality.

1. INTRODUCTION There has been a continuous increase in investment by various governmental and non-governmental organizations in new information technology and tools for effective operation and management. The U.S. Department of Commerce, Bureau of Economic Analysis state that as much as 50% of all new capital investment is made on information technology (IT) [1]. Consequently, Electronic Library Systems (ELS) has become an inevitable part of today’s educational system; a qualitative library system can directly or indirectly improve the quality of education. ELS aims to acquire, store, organize and preserve information for easy access and retrieval. Leedy [2] found out that information seekers often need the assistance of a Librarian, especially when the catalogues and guides were not useful. In recognition of this, many attempts have been made towards the establishment and improvement of the structure of the library to achieve high degree of usefulness, easier access and retrieval of information. This advancement gave birth to the concept of Electronic Library System (ELS) (also

referred to as Digital Library). Many Universities have digitized their library systems. However, while many resources have been devoted to developing these systems, library researchers have observed that digital libraries remain underutilized [3] and if these systems are not widely utilized, it will be difficult to obtain corresponding return on investments. Therefore, there is a clear need to identify and compare factors that can affect ELS acceptance and use by people from different cohort group especially people of different nationality and gender so that information system designers, school managers, library managers and others can formulate strategies to design systems that can be acceptable by all (males, females, international, national). It has been shown that women are significantly less receptive to IT than men and it is also agreed that understanding the gender differences on the acceptance of information technology will help in improving the overall quality of IS. Therefore, understanding the relationships between gender related constructs like nationality will be helpful in explaining why female students have lower technology acceptance.

68





The importance of this type of research is further underscored by United Nations’ recommendations: “...research institutions should, as appropriate, promote research on the interrelationship between... gender and age” [4]. As a result, the adoption and use of ELS in educational and research institutes has become a topic of broad interest to researchers and practitioners in management. Morris et al [5] found that the effects of gender on individual adoption and use of technology differ based on age. Specifically, they found that gender differences in technology perceptions is more pronounced among older workers while there is unisex pattern of results among younger workers and Wang et al [6] found that gender moderated the effect of performance expectancy, effort expectancy and self efficacy. In addition, the work by Venkatesh et al. [7] examined the simultaneous effects of age and gender. In this paper however, we argue that the effect exerted by gender on the independent constructs will vary depending on whether the user is a national or international student. International males will exhibit difference usage behavior from national males and likewise international females from national females. Consequently, we developed a NUTAUT adapted from UTAUT Model to account for the simultaneous moderating effect of gender and Nationality on UTAUT model. This was motivated by the observation that difference cohort groups of users (national and international students) exhibit different use behavior towards ELS based on not only gender but on nationality. Our research model not only predicts the varying degrees of acceptance for each user groups but also shows the degree of importance of each independent construct in determining acceptance for each group. The study significantly contributes towards understanding the acceptance of ELS in Academic environments and can also be useful to school managers, bank managers and other IS designers that makes decision about IS that is used by people of different nationalities.

There have been significant advances in the technical development of digital libraries in areas such as information storage, information retrieval, and system integration, resulting in dramatic improvements in their performance. While many resources have been devoted to developing these systems, library researchers have observed that digital libraries remain underutilized [3]. Digital libraries have received a lot of attention from researchers. Neuman [8] in her naturalistic inquiry detailed some of the difficulties 92 high school freshmen and sophomore displayed as they interacted with digital library databases. Her data revealed “basic difference between structures inherent in database and the conceptual structure that students bring to searching-differences so compelling that they seriously hamper students’ independent use of these resources” (p.74). The students’ lack of understanding of the organization of information hampered them from accessing the appropriate information for their research. The study demonstrated that information search has not become easier with the advent of technology.

The rest of this paper is organized as follows: In Section 2 we discuss the theoretical background of the study while Section 3 presents our NUTAUT model. Section 4 highlights methods employed in our research and Section 5 presents the analysis of the result. Section 6 discusses our finding followed by conclusions, limitations and recommendations for future work. 2. THEORETICAL BACKGROUND AND RELATED WORK

This section provides the theoretical background and related work in the area of digital library system, technology acceptance theories, gender in technology acceptance and UTAUT model.

2.1 Digital Library System

2.2 Technology Acceptance Theories

It is a common belief that introducing a new technology automatically results in service acceptance. However, several research findings dispute this claim, showing that there are several other factors that affect technology acceptance [9]. Many IS researchers have investigated various theories that could explain the acceptance of information technology. These theories include; the technology acceptance model (TAM) by Davis [10]; the theory of reasoned action (TRA) by Fishbein and Ajzen [11]; the theory of planned behavior (TPB) by Ajzen, [12]. The TAM model is the most widely used and has “perceived usefulness” and “perceived ease of use.” as its main elements. The model suggests that when users are presented with technology, “perceived usefulness” and “perceived ease of use” influence their decisions about how and when they will use the technology. The Perceived Usefulness is defined as “the degree to which a person believes that using a particular system would enhance his or her job performance,” while Perceived Ease of Use is defined as “the degree to which a person believes that using a particular system would be free of effort” [10]. We approached the adoption and discovery of critical factors that affect adoption of ELS from the perspective of technology acceptance. 2.3 Gender and Nationality in Acceptance and Usage of Technology

Gender can be defined as “the way members of the two sexes are perceived, evaluated, and expected to behave.”

69




[13]. Gender is a significant variable in explaining differential outcome in consumer behaviour research [14, 15]. There has been a limited amount of gender-based study in information technology research [16, 17, 18]. However, the differences between men and women have been studied in various contexts such as electronic mail [19], information retrieval [18], e-learning [20], communication technologies [21] and online purchasing behaviour [17]. Majority of the studies appear more favorably towards men than women. Anandarajan et al. [22] stated that men are more likely to access work pages than women. Nevertheless, gender was not linked with individual factors such as ease of use, frequency of use and time usage. Gefen and Straub [16] found that the perceptions of men and women vary. The perceived social usefulness of email was found to be lower with males than females. Conversely, men perceived ease of use higher than women. Nonetheless, the actual use of email did not vary across gender. Venkatesh et al. [18] proposed that gender would moderate the association between perceived usefulness, perceived ease of use, and subjective norm on intention to use the technology. These factors were more important for men than women [17]. Moreover, owning to the changes in the makeup and diversity of the workforce, gender differences are becoming increasingly important in managing the development and implementation of new technology ([19], [23]). As many have noted, gender and age are among the most fundamental groups to which individuals can belong and membership of such groups is likely to have a profound influence on individual perceptions, attitudes, and performance [24].

2.4 The Unified Theory of Acceptance and Use of Technology (UTAUT) Model

Attempts to develop a model with correct and high prediction of technology acceptance gave rise to as many as eight models, which have received support in recent literature. These models use different determinants in investigating acceptance of technology. A comparison of the determinants found in major acceptance and use models is presented in [25, 37, 18, 7]. UTAUT model, as shown in Figure 1, resulted from a study by Venkatesh [7] who systematically compared the eight previous models and their predictive factors. Determinants of acceptance in UTAUT are: performance expectancy (PE), effort expectancy (EE), social influence (SI) and facilitating conditions (FC) and the model also integrates four moderating factors (gender, age, Experience, and voluntariness) having varying influences on the primary constructs. The definition of the UTAUT constructs and moderators are given in Table 1 and 2 respectively. The combinations of the constructs and moderating factors have increased the predictive efficiency of acceptance to 70% this is a major improvement over the previous TAM model rates (35%) (Venkatesh et al. [7]).

s

Table 1: UTAUT Components

Table 2: UTAUT Moderators

Moderator Description Gender Gender roles have a strong psychological

basis and are enduring. Age Age has an effect on attitudes. Experience Deals with how long the user has used the

system. Voluntariness of use

If usage is voluntary or mandated.

3. MODEL FORMULATION

This section summarizes the NUTAUT and proposed the hypothesis that guides this study.

Determinant Description

Performance expectancy (PE)

Degree to which an individual believes that using the system will help attain gains in job performance.

Effort expectancy (EE)

The degree of ease associated with the use of the system.

Social influence (SI)

The degree to which an individual perceives that important others believe he or she should use the new system.

Facilitating conditions (FC)

The degree to which an individual believes that an organizational and technical infrastructure exists to support use of the system.

Behavioral Intention (BI)

The measure of the likelihood of an individual to employ the application.

Use Behavior (UB)

This measures the acceptance of the technology.

Figure 1: UTAUT Model Venkatesh et al., [26]

70




3.1 Nationality Based UTAUT (NUTAUT)

Since the development of UTAUT model, it has attracted attention of many scholars in IS research because of its predictive efficiency of 70% this is a major improvement over the widely used TAM model with predictive capacity of 35% [26, 18]. The scholars [27, 28] validated the model and others [26, 30, 37, 31] extended it in different contexts, including multicultural studies [32], and all found its constructs highly predictive [7]. This, in addition to the fact that the moderating variables offer

flexibility to allow the introduction of new dimensions into the model, was the major motivation for the use of UTAUT model in the current investigation. NUTAUT was adapted from UTAUT by introducing a new modulating variable nationality as shown in Figure 2. Nationality was introduced based on the assumption that the UTAUT independent variables PE, EE, SI will impact on BI differently and BI with FC will also impact on UB differently when moderated by nationality. The definition of these construct are given in Table 1 and Table 2.

PE

3.2 Hypotheses

The expectations are that the survey will provide evidence of varying degree of acceptance by male and female students and also prove that the independent variables will affect acceptance at different degrees for these groups. Four hypotheses have been postulated to guide this study. It has been found that gender differences moderate the effect of independent constructs on technology acceptance and that it is more important for males [5, 33, 34, 35], we expect that performance expectancy related to technology usefulness and its influence on acceptance will be more important for international and national males than their female counterpart, therefore, resulting in greater influence on behavior intention in predicting ELS adoption. Again, Examining social influence from a gender perspective, the literature on gender differences and gender roles suggests that women have higher affiliation needs and are thus more concerned with pleasing others and more likely to conform to majority opinions ([35], [36]) while gender role theory suggests that women tend to value and

respond to opinions of their social group. Following from these we formulated the following hypothesis:

H1: Gender and Nationality will moderate the effect exerted on behavior intention by social influence in a way that the importance of social influence will be more pronounced for males and females of international origin. . H2: Gender and Nationality will moderate the relationship between performance expectancy and behavioral intention such that the importance of performance expectancy will be more pronounced within the national group when compared to their international counterpart H3: Gender and Nationality will moderate the effect exerted on behavior intention by effort expectancy such that the influence of effort expectancy will be more pronounced for males and females of international origin than their national counterpart. H4: Facilitating condition will be a more important determinant for international males and females than their national counterpart.

Gender Experience Voluntariness Nationality

FC

SI

EE BI UB

Age

Figure 2: Nationality Based UTAUT (NUTAUT)

71




4. METHODS

4.1 Survey Instrument The data-gathering instrument used for this study was a self-administered online questionnaire. The questionnaire was based on the pre-existing tool developed by Venkatesh [7] and has been used by Anderson and Schwager [27], Moran [37] and Tibenderana & Ogao [31]. The research question was divided into three sections. Section 1 contained 18 close-ended questions which collected the participant demographic information and their experience with computers and ELS. Section 2 contained 21 questions about ELS hardware and services provided by Middle East Technical University (METU) library. These questions collected the students’ awareness of these facilities and services. The respondents chose either a “Yes”, “No” or “Not Sure” answer in response to each ELS services and facilities indicated. Section 3 contained 25 questions with a 5-point Likert scale where 1 represented ‘strongly agree’ and a 5 represented ‘strongly disagree’.

4.2 Participants

There were a total of 116 participants: 56 Males (28 international and 28 national) and 60 Females (30 international and 30 national) graduate students. The participants were distributed across various departments and the number of international graduates and National graduate students from each departments and schools were fairly evenly distributed to ensure valid comparison. Moreover, the number of male and female participants was also fairly evenly distributed. A pilot study was carried out on 10 participants (4 international graduate students and 6 National graduate students) to ensure the clarity of the questions for the respondents and to eliminate problems that might arise while recording and analyzing the data. The results from the respondents were analyzed to ensure that it could answer the Hypothesis. The survey was confirmed adequate for the research; although some minor rephrasing was made on some questions to increase understandability.

4.3 Electronic Library Services Offered in the University. Eleven ELS services were listed in the questionnaire and respondents were asked questions about their availability. Again more than 70% of all the respondents confirmed the availability of all the listed services with the exception of Electronic Database which has only 8% of the entire respondent confirming the availability. This also indicates a high degree of awareness

of the services apart from Electronic Database which needs adequate sensitization.

5 RESULT ANALYSIS

5.1 Validation of the survey instrument and NUTAUT

The data analysis was done using SPSS 17 and Linear Structural Relations (LISREL) structural equation modeling tool. SPSS 17 was adopted to conduct Principal Components Analysis (PCA) and to assess the validity of the scale. The Cronbachs’ Alpha was calculated to examine the reliability of each factor, the Alpha values of the questionnaire exceeded 0.8 (Table 3, column 5), demonstrating the good reliability. Before conducting PCA, Kaiser-Meyer-Olkin (KMO) and Bartlett sphericity test was checked to measure for the sampling adequacy [38]. The KMO were all >0.700 and the result of Bartlett sphericity test was significant at <0.001 (Table 3, column 3 and 4). Thus data were suitable to conduct factor analysis [11]. The factor loadings and the corresponding factor scores (weights) for each variable were generated. The dimensionalities of the twenty-five statement making up the six constructs of the instrument were each subjected to factor analysis resulting in the removal of two questions and the models re-estimated. Each factor has larger loading on its corresponding factor (>0.7) than cross-loadings on other factors (<0.4). Thus these items could effectively reflect factors since they had good validity including convergent validity and discriminant validity [40]. Structural Equation Model software LISREL on the other hand was employed to estimate the path coefficients and to validate and test models hypotheses. We used Confirmatory Factor Analysis (CFA) to test for the model fitness on the data. The results show that the hypothesized model is recursive, uni-directional (Table 3 & 5). The fit indices of the model are listed in Table 4 and Table 5. The tables list the recommended value and actual value of each fit index, the actual value was better than the recommended value. Thus the model was a good fit to the data. Table 5 summarizes the results of the t-test analysis which further confirms the validity of the models. Also, as shown in the Table 6 all the standardized loadings of items on their corresponding factor were larger than 0.7 further proving good convergent validity [41].

5.2 Availability of ICT Hardware in the Library

In response to the questions asked about the availability of eight ICTs’ hardware in the Middle East Technical University (METU) library, more than 85% of all the respondents confirmed that computer, printers, internet, bar code readers, CD-ROM readers/writers, security check systems and photocopying machine are available in the library. This indicates that students are aware of the ICT

72




hardware. In response to the questions about the availability of 11 ELS services listed in the questionnaire, more than 70% of all the respondents confirmed the availability of all the listed services with the exception of Electronic Database with only 8% of the entire respondent confirming its availability. This also indicates a high degree of awareness of the hardware apart from Electronic Database that needs adequate sensitization.

5.3 International versus National Students

The four groups’ result estimation using LISREL are as shown in Figures 3, 4, 5 and 6. The comparison of the results from the two analyses is summarized in Table 7 and 8. The contributions of the various independent constructs PE, EE FC and SI are shown in Table 7 column 3 and 4. It shows that FC is the most important predictor of acceptance for the two groups. The international female students model shows predictive efficiency of 32% and 56% for the dependent construct of behavioral intent and use behavior (a total predictive capability of 88% for the dependent variables) as shown in Figure 3 and Table 8, the national female students’ model shows predictive efficiency of 20% and 62% (a total predictive capability of 82% for the dependent variable) as shown in Figure 4 and Table 8, international male students’ model shows predictive efficiency of 25% and 61% for the dependent construct of behavioral intent and use behavior (a total predictive capability of 86% for the dependent variables) as shown in Figure 5 and Table 8. The national male students’ model shows predictive efficiency of 18% and 78% (a total predictive capability of 96% for the dependent variables) as shown in Figure 6 and Table 8. This means that the four groups accept and use ELS though at varyingt degrees.

Table 3: Construct Validity and Reliability Measure

Group Comparison KMO

Bartlett Sphericity

Cronbach’s Alpha

P-Value

Chi-Square/df

RAMSEA

Recommended Value >0.5 <0.05 ≥ 0.7 <0.05 <3 ≤0.08

International 0.758 000 0.879 0.0000

0 1.51 0.037 Female

National Students 0.702 000 0.921 0.0000

4 1.56 0.071

International Students

0.730 000 0.883 0.00004 1.56 0.071

Male National Students 0.740 000 0.962 0.0000

4 1.56 0.071

Table 4: Goodness-of-Fit Results of the LISREL General Model

(Note: x2/df the ratio between Chi-square and degrees of freedom, GFI is Goodness of Fit Index, AGFI is Adjusted Goodness of Fit Index, CFI is Comparative Fit Index, NFI is Normed Fit Index, RMSEA is Root Mean Square Error of Approximation) Table 5: T Test and P Values for Participant Groups

Dependent Variable

Category T-test P-Value

National 2.41 <0.001 Behavioral Intention International 3.25 <0.001

National 4.30 <0.001

Female

Use Behavioral International 4.20 <0.001

National 2.20 <0.001 Behavioral Intention International 4.21 <0.001

National 4.37 <0.001

Male

Use Behavioral International 5.40 <0.001

Table 6: The factor-loading matrix with varimax rotation

Factor Loading Matrix with Varimax rotation for All Students

FC EE PE

SI BI UB PE1 .245 .068 .813 .253 .159 -.039

PE2 .219 .180 .779 .033 .084 .217

PE3. .232 .084 .815 .176 .075 .282

PE4 .155 -.055 .641 .325 .167 .300

EE1 .105 .707 .070 .076 .280 .135

EE2 .491 .605 .194 .008 -.019 .244

EE3 .230 .787 .293 -.006 .003 .006

EE4 .115 .767 .025 .312 .167 .191

EE5 .249 .811 .106 .164 .073 -.015

EE6 .171 .810 .102 .035 .190 .047

SI1 .140 .133 .031 .825 .074 .179

SI2 .157 .271 .160 .801 .098 -.095

SI3 .214 .152 .132 .792 .072 .045

FC1 .768 -.029 .364 .180 .063 .180

FC2 .673 .329 .112 .035 .354 .038

FC3 .828 .100 .125 .067 -.016 .202

FC4 .758 .069 .138 -.019 .144 .200

FC5 .730 .318 .154 .085 .306 .-129

BI1 .171 .197 .124 .026 .905 -.002

BI2 .125 .054 .126 .160 .895 .142

UB1 .304 .410 .059 .104 .149 .711

UB2 .038 .481 .348 .042 .055 .723

UB3 .366 .176 .160 .126 .067 .763

Extraction Method: Principal Component Analysis.

Rotation Method: Varimax with Kaiser Normalization.

a. Rotation converged in 6 iterations.

GFI AGFI RMSEA χ2/df P-value CFI NFI

Recommended >0.90 >0.80 <0.08 <3 <0.05 >0.90 >0.90

Actual Value 0.956 0.860 0.069 1.45 <0.0001 0.974 0.943

73




Figure 6: National Male Participants’ Model

Figure 5: International Male Participants Model Figure 3: International Female Participants Model

Figure 4: National Female Participants Model

74




Table 7: Comparison of National and International Model Contributions

Table 8: Comparison of International and National Models

Prediction

6. DISCUSSIONS AND CONCLUSIONS

This study reveals a rich set of pattern and results. Due

to space limitations, we will only highlight the most important findings without going into details on specific results. In general, the results suggest that Nationality and Gender considered simultaneously is a significant moderator that affects the acceptance of ELS and the effect exerted by UTAUT constructs on individual groups (International males, International females, National males and National female students) vary considerably. Facilitating condition significantly influences the acceptance and use of ELS for the four groups.

In general, the NUTAUT does predict the successful

acceptance of ELS by graduate though at varying degrees Table 8. The participants showed higher inclination to use ELS by showing higher percentage of use behavioral. The researcher believes that this is as a result of the role played by readily available access and easy to use library facility in-campus than out-campus since majority of the participants stays inside the university campus and can

easily and conveniently access the ELS than the students living outside the campus. To actually increase acceptance of ELS, the research suggest that the university and any other university that is working towards increasing acceptance of ELS should make accessibility (facilitating conditions and effort expectancy) of the ELS at both outside and inside the campus easy.

H1: Gender and Nationality will moderate the effect exerted on behavior intention by social influence in a way that the importance of social influence will be more pronounced for males and females in the international category.

This hypothesis is supported. Though previous works

on gender roles suggested that women have higher affiliation needs and are thus more concerned with pleasing others and more likely to conform to majority opinions ([35], [36]), the result of our analysis in fact

shows that social influence has a more pronounced effect on international students (both males and females) than their national counterpart by showing that social influence predicts 28% for international females, 12% for national females, 20% for international males and 2% for national males as shown in Figure 3, 4, 5 and 6 and also summarized Table 7. This is as expected since international students are likely to move in group and, therefore, are easily influenced to use this system by people that matters to them in the environment. This can, to an extent, be likened to the effect of facilitating condition; through social facilitation and social comparison the groups function as a motivator and can provide necessary assistance especially for does that are new to the system. . Practically, this study results suggest that organizations should use different strategies in motivating the use of a new technology for different situations/groups. For some information systems whose usage is mandatory, those factors contributing to social influence such as the instructors/supervisor's prodding might work. However, when the usage is voluntary just like the case of ELS, the managers might want to think of better ways to promote usage probably through social facilitation and social comparison. As a matter of fact, though social influence is a significant factor for international students, the effect can greatly be reduced by training and probably experience and therefore should not be used as a strong motivator. This is consistent with the previous studies (e.g., Venkatesh & Davis [7]); the effect of subjective norm to usage intention is significant under mandatory and inexperienced use situation.

Groups Constructs International (N=58)

National (N=58)

No.of Questions

Asked

No.of Questions Retained

PE 0.20 0.30 4 4 EE 0.48 0.58 6 5 SI 0.28 0.12 3 2 FC 0.78 0.63 6 2 BI 0.32 0.20 3 2

Female (N=60)

UB 3 2 PE 0.41 0.50 4 4 EE 0.35 0.51 6 5 SI 0.20 0.02 3 2 FC 0.74 0.60 6 2 BI 0.25 0.18 3 2

Male (N=56)

UB 3 2

Gender Model Behavioral Intention

Use Behavior

Total

International 32% 56% 88% Female

National 20% 62% 82%

International 25% 61% 86% Male

National 18% 78% 96%

H2: Gender and Nationality will moderate the

relationship between performance expectancy and behavioral intention to use technology, such that the importance of performance expectancy will be more

75




pronounced within the national group when compared to their international counterpart.

This Hypothesis is only partially supported. There is a

significant difference in the effect exerted on behavior intention by performance expectancy between the international and national group. The model shows that national male group has performance expectancy of 50% as opposed to 41% of international male group while national females have performance expectancy of 30% as opposed to the 20% of the international female counterpart as shown in Figure 3, 4, 5 and 6 and also summarized Table 7. Though the performance expectancy of national male group is higher at 50%, the performance expectancy of the national females (30%) is lower than performance expectancy of international males (41%). This means that within the gender groups, performance expectancy is a significant factor for male than female. This shows that males have higher tendency to use the ELS system as long as it is useful. In other words, performance expectancy is an important factor for males despite their nationality while effort expectancy determines acceptance for females more than performance expectancy despite nationality. This finding follows the work of [33, 34, 35].

More importantly, this result reveals that perceived

performance expectancy of a system does vary based on gender but not necessary based on nationality though the degree of contribution still differs based on nationality of the students. The most plausible explanation is that performance expectancy cannot be interpreted in isolation without some other factors like effort expectancy and facilitating condition; the usefulness of any system might never get to be appreciated /discovered if the system is so complex that end users hardly use it. Most international users (males and females) lack the technical knowledge needed to handle the complexity of the ELS especially at the early stage of their arrival. This is understandable since difference policies guide the design and implementation of ELS in difference countries. Therefore training of new students especially the international students on the use of ELS on arrival will surely increase acceptance.

H3: Gender and Nationality will moderate the effect exerted on behavior intention by effort expectancy such that the influence will be more pronounced for males and females of national category than their international counterpart.

This is supported. Effort expectancy contributes 51% and 58% for national males and females respectively as opposed to 35% and 48% for international males and females respectively as shown in Figure 3, 4, 5 and 6 and also summarized Table 7. This means that for national

students (both males and females) acceptance is more dependent on effort expectancy than international students (both males and females). This is partially contrarily to some previous work [33, 34, 35] that shows that males have higher tendency to use the ELS system even when it is complex. In other words, Performance expectancy is an important factor for males more than the effort expectancy while effort expectancy determines acceptance for the females more than performance expectancy. Examining gender and nationality simultaneously shows that both females and males in the national group attach more importance to the expected effort required to use the system. In other words national students are unlikely to use the system if it is difficult to use even if it is useful. This can also be explained by the fact that availability of alternatives or competition generates negative effect affecting perceived effort expectancy. The national students have other sources of getting materials for their research easily than international students and might not afford to spend a lot of time and energy searching through complex ELS while international students can afford to spend an extra effort to use the system. This could also mean that in the absence of alternative, performance expectancy might become as important as effort expectancy. This also agrees with the work of Pontiggia and Virili [42] which shows that technology acceptance is basically a choice among different alternative technologies/tools to accomplish user tasks. H4: Facilitating condition will be a more important determinant for international males and females than their national counterpart.

This hypothesis is again supported with 74% and 78% prediction for international males and females respectively, as opposed to 60% and 63% for national males and females respectively as shown in Figure 3, 4, 5 and 6 and also summarized Table 7. Though international students group (males and females) shows high contribution from facilitating condition than their national counterpart, in general, facilitating condition has the highest contribution to acceptance than any other variables in all the groups. This means that irrespective of the group, facilitating condition is very crucial. Our initial assertion that facilitating condition will not be as important for national students owing to the availability of alternatives which includes resources from friends and families seems to be wrong. This can possibly be explained by the fact that the University invests considerable amount of resources to provide both online, offline and remote access to the ELS, so international students as well as national students still exhibits use behavior on the ELS despite availability of alternatives. This is consistent with the empirical studies of Thompson et al [43] that found the direct effect of facilitating condition on usage behavior and some others [ 7, 18].

76




In general, the NUTAUT does predict the successful

acceptance of ELS by international males, national males, international females, and national females with a total predictive capacity of 86%, 96%, 88% and 82% respectively. The groups showed higher inclination to use ELS by showing higher percentage of use behavioral. The researcher believes that this is as a result of the role played by readily available access and easy to use library facility in-campus than out-campus since majority of the participants stays inside the university campus and can easily and conveniently access the ELS than the students living outside the campus. To actually increase acceptance of ELS, the research suggest that the university and any other university that is working towards increasing acceptance of ELS should make accessibility (facilitating conditions and effort expectancy) of the ELS at both outside and inside the campus easy.

This study shows that the variables, facilitating condition, effort expectancy, performance expectancy and social influence which are the critical components that affect acceptance and use of ELS exert varying effects on acceptance not just depending on gender but also on nationality of the users. Facilitating condition remains the most important factors for the four group of user, however, for international female group effort expectancy, social influence and performance expectancy listed in decreasing order of importance are the critical factors, and for the national female group listed in decreases order of importance effort expectancy, performance expectancy and social influence are the critical factors. Likewise for international males, listed in decreasing order of importance, performance expectancy, effort expectancy, social influence are the critical factors that effects acceptance while for national males effort expectancy, performance expectancy and social influence listed in decreasing order of importance are the determinants of acceptance. Though this research focused on the acceptance of ELS in educational institutions, the conclusions derived here could have significant implications for organizations, business and future researchers. From the researchers’ point of view, most studies carried out in recent years have treated technology users in the same way regardless of their gender and nationality. Nevertheless, the effect of this factor on acceptance has being found significant. These variations found in users perceptions in relation to the use of ELS illustrate the need for considering the effect of gender and nationality in the design and improvement of these systems. It is concluded that the differences discovered in this research are not all similar to those previously established by studies investigating only gender or gender and some other variables like age. Males in general demonstrated a higher acceptance of ELS than females.

In conclusion, organizations, managers, designers and developers can raise intention to use ELS through facilitating conditions, , effort expectancy and performance expectancy. Institutions may arrange training programs to develop knowledge about the benefits and usage of ELS system compared to some other randomized searches. The training should be included as part of the event to feature during the induction of new students especially for international students. Also paring international students to pair leaders could also function as a social facilitation to motivate them to use ELS.

This research was carried out in a university

environment and may not reflect ELS acceptance outside the university environment, although we plan to validate NUTAUT elsewhere. In the researchers’ opinion, 116 participants are limited number for this type of study therefore proposes conducting similar study on larger number of participants and in another environment.

ACKNOWLEDGEMENTS

The work reported in this paper was carried out when the Author was a graduate student of Middle East Technical University, Turkey. The author appreciates their provision of enabling environment for which the research was conducted. REFERENCES [1] J. C. Westland and T. Clark (2000) Global

Electronic Commerce: Theory and Case Studies. Boston, MA: MIT Press. [2] Leedy, P. (1993). Practical Research, Planning and

Design: Columbus, Ohio, Merrill, 17-42. [3] Wood, F. et al, (1995). Information Skills for Student

Centered learning. The Association for Information Management. Milton Keynes, London, UK, 134–148.

[4] “Gender and Ageing: Problems, Perceptions, and Policies,” United Nations Economic and Social Council, New York, E/CN.6/1999.3, 1999.

[5] Morris et al (2005): Gender And Age Differences In Employee Decisions About New Technology. Ieee Transactions On Engineering Management, Vol. 52, No. 1,

[6] Wang et al (2010), Mobile Activities in Mobile Internet.International Wireless Communication and Mobile Computing: p 1282-1284.

[7] Venkatesh V. et al (2003) “User acceptance of information technology: Toward a unified view,” MIS Quart., vol. 26, pp. 425–478,.

[8] Neuman, D. 2004. Learning and the Digital Library. Youth information seeking behavior. Theory, model and issues (pp. 65-95).

77




[9] Carlsson, C., J. et al (2006). Adoption of Mobile Device/Services Searching for Answers with the UTAUT. Proceeding of the 39th Hawaii International Conference on System Science. Hawaii. USA. 1 – 10.

[10] Davis, F. D. et al (1989). User Acceptance of Computer Technology: A Comparison of Two Theoretical Models. Management Science. 35(8): 982-1002.

[11] Fishbein, M. and I. Ajzen, (1975). Attitude Intention and Behavior: An Introduction to Theory and Research. Reading, MA: Addison-Wesley Publishing Company. [12] Ajzen, I. (1991). The Theory of Planned Behavior. Organizational Behavior and Human Decision Processes, 50,179-21 [13] Feingold (1993), “Cognitive gender differences: A developmental perspective,” Sex Roles, vol. 29, no. 1-2, p. 91 [14] Forsythe, S. and L. Chun. (2000) Exploring Gender

Differences in Online Behavior, in International Textile and Apparel Association (ITAA): Cincinnati, Ohio.

[15] Qualls, W. (1987) Household decision behavior: The impact of husbands’’ and wives sex role orientation. Journal of Consumer Research,. 14: p. 264-279.

[16] Gefen, D. and D. Straub. (1997). Gender differences in the perception and use of e-mail: An extension to the Technology Acceptance Model. MIS Quarterly, 21(4): p. 389- 400.

[17] Thomas, P. and K. Taskov (2007), Extending gender differences and technology acceptance to a database environment in In proceedings of the 6th annual ISOnEworld conference. Las Vegas, Nevada, USA.

[18] Venkatesh, V. and M. Morris. (2000), Why don't men ever stop to ask for directions? Gender, social influence, and their role in technology acceptance and usage behaviour. MIS Quarterly, 24(1): p. 115-39.

[19] M. K. Ahuja (2002), “Women in the information technology profession: A literature review, synthesis and research agenda,” Eur. J. Inf. Syst., vol. 11, pp. 20–34,.

[20] Ong, C. and J. Lai (2006), Gender differences in perceptions and relationships among dominants of e-learning acceptance. Computers in Human Behavior, 22(5): p. 816-829.

[21] Ilie, V., et al. (2005), Gender Differences in Perceptions and Use of Communication Technologies: A Diffusion of Innovation Approach. Information Resources Management Journal. 18(3): p. 13-31.

[22] Anandarajan, M., C. (2000) Simmers, and M. Igbaria, An exploratory investigation of the antecedent and impact of internet usage: an individual perspective. Behaviour & information technology. 19(1): p. 69-85.

[23] R. Woodfield, (2002) “Women and information systems development: Not just a pretty interface,” Inf.

Technol. People, vol. 15, pp. 119–138,. [24] B. A. Nosek et al (2002) “Havesting implicit group

attitudes and beliefs from a demonstration website,” Group Dynamics: Theory, Research, Practice, vol. 6, pp. 101–115,.

[25] Lee, Y. et al (2003). The Technology Acceptance Model: Past, Present, and Future. Communications of the Association for Information Systems. 12(50): pp.752-780

[26] Meister, D. B. and Compeau, D. R. (2002). Infusion of Innovation Adoption: An Individual Perspective: Proceedings of the ASAC. Winnipeg, Manitoba.

[27] Anderson, J. E. and Schwager, P. H. (2006). The Tablet PC: Applying the UTAUT Model, Paper presented at the American Conference on Information Systems, Acapulco,Mexico.

[28] Louho, R., M. et al (2006). Factors Affecting the Use of Hybrid Media Applications. Graphic Arts in Finland. 35: 3.

[29] Cody-Allen, E. and Kishore, R. (2006). An Extension of the UTAUT Model with E- Eqality, Trust & Satisfaction Constructs. CPR: 183-189.

[30] Heerik, M. et al (2006). Human-Robot User Studies in Eldercare: Lessons Learned. Institute for Information Engineering, University of Amsterdam, Oudweg, The Netherlands.

[31] Tibenderana, P. K. and Patrick J. O. (2008). Acceptance and Use of Electronic Library Services in Uganda Universities. ACM JCDL.

[32] Oshlyansky, L. et al (2007). Validating the Unified Theory of Acceptance and Use of Technology (UTAUT) tool cross-culturally. Proc. of HCI. 2 BCS, 83-86.

[33] J. M. Arceneaux, G. M. et al (1996), “Gender differences in WAIS-R age-corrected scaled scores,” Perceptual Motor Skills, vol. 83, no. 3, pp. 1211–1216

[34] R. C. Barnett and N. L. Marshall, (1991) “The relationship between women’s work and family roles and their subjective well-being and psychological distress,” in Women, Work and Health: Stress and Opportunities, M. Frankenhaeuser, V. Lundberg, and M. A. Chesney, Eds. New York: Plenum, pp. 111–136.

[35] A. H. Eagly, (1987) “Reporting sex differences,” Amer. Psychol., vol. 42, pp. 756–757.

[36] J. B. Miller. (1986) .Toward a New Psychology of Women. Boston, MA: Beacon, 1986.

[37] Moran, M. J. (2006). College Student’s Acceptance of Tablet PCs and Application of the Unified Theory of Acceptance Technology (UTAUT) Model. Ph. D. Thesis. Capella University. Mennesota, USA.

[38] Kaiser, H. (1970). A Second Generation Little Jiffy. Psycometrika, 35, 401-415.

[39] Guo Z. (1999). Social Statistics Analytical Methods—the Application of SPSS. Beijing: China Renmin

78




University Press [40] Gefen D et al (2000). "Structural Equation Modeling

and Regression:Guidelines for Research Practice," Communications of the Association for Information Systems, vol 4, pp. 1-70.

[41] Bagozzi R P and Yi Y. (1988). "On the evaluation of structural equation models," Journal of the Academy of Marketing Science, 16(1), pp. 74-94.

[42] Pontiggia A. V. and Virili, F. (2009). Network effects in technology acceptance: Laboratory experimental

evidence. International Journal of Information Management, 30(1): pp. 68 – 77.

[43] Thompson R. L. et al (1991). Personal Computing: Toward a Conceptual Model of Utilization. MIS Q. 15(1):125–43.

[44] Al-harby, et al (2009). The effects of gender differences in the acceptance of biometrics authentication systems within online transaction. International Conference on CyberWorlds. pp. 203 – 210.

79

VOL. 1, NO. 2, Oct 2010 E-ISSN 2218-6301




80

Encouraging the Inclusion of Evolutionary Psychology into Information Systems’ Theoretical Repertoire as an Emerging Trend

Chon Abraham

Mason School of Business The College of William and Mary

P.O. Box 8795 Williamsburg, VA 23187, USA

Email: [email protected]

ABSTRACT Social and cognitive psychology frames have dominated IS researchers’ attention in developing and applying theory to explain human behavior in the context of IS. However with a few exceptions, the information systems literature has largely ignored the possible explanatory power of evolutionary psychology. This is a major omission, particularly as Darwin argued that perception and cognition, areas of IS interest, are the product of evolution of the human brain adapting to motivate behavior that supports productive decision-making. It seems appropriate for IS research to recognize that humans are an evolved, social, cognitive species. In this paper, we encourage IS scholars to appreciate the potential contribution of evolutionary psychology as an emerging trend to enrich the theoretical repertoire for information systems.

Keywords: evolutionary psychology, technology acceptance, evolved psychological mechanisms, human decision-making.

1. INTRODUCTION “It is the theory which decides what can be

observed.” (Einstein in a conversation with Heisenberg [1])

Social and cognitive psychology frames have dominated IS researchers’ attention in developing and applying theory to explain human behavior in the context of IS. However with some exceptions [2],[3],[4],[5],[6],[7],[8], the information systems literature has largely ignored the possible explanatory power of evolutionary psychology. A most recent paper urges the IS research community to rectify this major omission by asserting that “evolutionary information systems theories will have to be integrated with non-evolutionary theories to fully explain certain information systems phenomena” [8]. In this paper, our perspective is that IS scholars have largely ignored the potential contribution of evolutionary psychology, we opine that this is a serious oversight, provide evidence to support this assertion, and encourage its inclusion to enrich our theoretical repertoire as an emerging trend. As much of our seminal theories in IS are based on social and cognitive psychological frames, it is prudent to continue exploring the underlying evolutionary perspective to give a more comprehensive understanding of human decision making regarding technology. With the acceptance of Kock’s (2009) paper, the door to discovering the potential of evolutionary psychology has been opened,

but getting scholars to walk through that door will still be problematic much in the same vein that is has taken over 150 years for our academic cousins in psychology to come to appreciate the influence of evolutionary based tents in shaping their discipline, as evident in the special issue dedicated to the Darwinian influence in both mainstream social and cognitive psychology in American Psychology [9]. Since many of the IS theoretical frames are rooted in mainstream social and cognitive psychology and as scholars we seek to constantly expand out theoretical repertoires to enrich our discipline, it is prudent to unveil the applicability of evolutionary tenets in IS research by recognizing much in the same vein as our academic cohorts in psychology, that humans are indeed social, cognitive as well as evolved species for which behavior, specifically decision making, is a manifestation of our adaption over time to promote survival of our species. Our perspective in this paper is that it is now time to unearth the influence of evolutionary theories in human decision making regarding technology, and we do so by introducing the IS research community to a particular categorization of the evolved psychological mechanisms (EPMs) (i.e., the innate motivators of behavior based on ancient encounters and problems that promoted ancestral survival and impact human motivation and behavior [10],[11],[12]. EPMs can be categorized as the four human drives (i.e., the drive to acquire, bond, comprehend, and defend [10],[13]. We also think it necessary to demonstrate how these human drives


VOL. 1, NO. 2, Oct 2010 E-ISSN 2218-6301




81

have already manifested in technology innovations unbeknownst to the vast majority of IS research community thus far. In doing so, we strengthen our theoretical IS repertoire with the knowledge of these drives that help explain more comprehensively the human decision making process with regards to technology. Thus, the trifecta of social, cognitive, and evolutionary tenets should be considered when continuing to study human behavior in technology acceptance. This paper is structured as follows. We first provide fundamental tenets of evolutionary psychology to further demonstrate its importance to understanding human behavior. Second, we then move to consider the Four-Drive model, which provides a high level abstraction of our many evolved psychological mechanisms. This Four-Drive model, by focusing on the fundamental evolutionary goals of each drive, provides scholars with a succinct means of understanding behavior in many contexts. Third, we illustrate how the four drives manifest themselves in IS settings to show their potential applicability to the IS domain. Finally, we discuss the implications of adding evolutionary psychology to the theoretical underpinnings of IS and suggest two research directions: a new path based on the neuroscience paradigm to explore evolved psychological mechanisms, and an augmentation of the dominant social science logic to include the four drives. 2. EVOLUTIONARY PSYCHOLOGY In 1975, Edward O. Wilson introduced his controversial thesis on sociobiology [14]. He proposed that human behavior was influenced by evolutionary development. Initially, his work was vigorously attacked, but ultimately his contention won the day [15]. Sociobiology, rebranded as evolutionary psychology [11], has entered mainstream thinking in some academic disciplines, yet it is very much on the periphery, essentially out of sight, of IS academic thought. Evolutionary psychology has asserted itself as a unifying theory that enfolds aspects of social and cognitive psychology with the tenets of evolutionary biology, cognitive science, anthropology, and neuroscience [11]. Human nature, or the “evolved, reliably developing, species-typical computational and neural architecture of the human mind and brain” [16], has gained recognition as a valuable lens for understanding how EPMs ultimately impact behavior. The central principle of evolutionary psychology is that humans, then and today, possess mechanisms that developed over time as instructions for behavior, and when we encounter a problem similar to that frequently

encountered by humankind’s forebears, we still act on that same set of mechanisms [17]. These mechanisms were encoded because of successful responses by our ancestors to recurring survival and reproduction challenges. Such behaviors included gaining status, forming social coalitions, protecting oneself, selecting a mate, appeasing curiosity, and recognizing danger, and the need for communication (i.e., language) [17],[18],[19],[20]. These evolved psychological mechanisms guided formulation of perceptions, problem resolution, and adaption to local environments [13],[17],[18],[19],[20]. For tens of thousands of years, humankind was exposed to relatively unvarying or slowly varying environmental conditions. New mechanisms developed when new and relatively stable situations were sustained in the environment for generations of humankind [17], [21]. Evolved psychological mechanisms have had sufficient buffering time to adapt to the environmental demands—until recently. These evolved mechanisms formed in our long evolution did not necessarily develop in coherence with each another. Instead, each developed in response to a particular adaptive problem and is thus functionally specialized [17],[22],[23]. Consequently, evolved psychological mechanisms coexist, but may compete on some dimensions. For example, the tendency to exhibit jealousy in mating situations counters the mechanism to form coalitions that ensure survival and reproduction. The behavioral manifestation of any given mechanism is not fixed, but depends heavily on the environment and the cues that motivate the behavior based on the influence of the dominant drive [24]. Nonetheless, humans are highly prone to exhibit certain patterns of behavior when faced with situations that closely resemble the ones that shaped our evolved psychological mechanisms [18]. The startle reaction when we see a snake unexpectedly is a good example of an innate mechanism at work [25]. From an evolutionary perspective, human nature and the environment are viewed as two inseparable forces [26]. Not only does the ecological environment influence a human’s behavior (as demonstrated in the snake example), but our actions are also determined by the presence of others in that environment [27]. “Human beings acquire their typical human psychological characteristics, powers and tendencies in ‘symbiotic’ interactions with other human beings” [28]. Thus, social exchange played a vital role in the formation of evolved mechanisms. Perhaps one of the most salient developments of sociality has been the creation of language [27]. Besides ecological and social influences, cognition was (and still is) a necessary ingredient in the calibration of evolved psychological mechanisms [16].

VOL. 1, NO. 2, Oct 2010 E-ISSN 2218-6301




82

Table 1: Drives and brain location

Cognition may be viewed as a short-term, phenomenon-driven activity that is necessary to analyze, prepare, and adjust evolved psychological mechanisms for long-term reproductive benefits [29]. Even though evolutionary psychology proposes that, depending on the situation, some underlying evolved psychological mechanisms may be activated [29], these mechanisms may be supported or overwritten by cognition. In summary, the combination of social, cognitive, and evolutionary psychology has the potential to explain more fully human behavior in a wide variety of situations, including many in which humans fund, analyze, design, implement, and use information systems. Evolutionary psychology has the potential to increase our understanding of IS phenomenon, so let us now examine how might incorporate it into the discipline.

THE FOUR-DRIVE MODEL As one of its fundamental tenets, evolutionary psychology asserts that our human architecture is an assembly of hundreds, and possibly thousands, of functionally specialized and dedicated modules, each designed to solve a particular adaptive problem [17],[30],[31]. This level of granularity does not work well for all situations, and as a starting point for research and practice, it is useful to have a higher-level abstraction. The Four-Drive model [10],[13] provides such a higher-level perspective by considering what these many modules seek to appease. It maintains that we have a set of four generic drives (i.e., the drive to acquire, bond, comprehend, and defend) that evolved in response to solving ancient adaptive problems and that still influence modern human behavior. The four drives might not be the only high-level human drives, but they are “central to a unified understanding of human behavior” [32]. These drives have been reported, implicitly or explicitly, by scholars for centuries [32]. Environmental inputs are evaluated in terms of their potential to satisfy or threaten each of the four drives. Thus, we would expect the drives to influence information systems use, just as they influence our everyday life. The drives are contended to be independent of one another, which is empirically supported by the observation that they are located in different parts of the brain (Table 1). Satisfying one drive does not satisfy another, and it is doubtful that any drive can be satiated.

Drive Definition Brain location

Acquire Drive to seek status, take control, gain objects and personal experiences that we value

Nucleus accumbens (limbic area) (Becerra 2001)

Bond Drive to form social relationships and develop mutual caring commitments with other humans

Hypothalamus and anterior thalamus (limbic area) (LeDoux 1996)

Comprehend Drive to collect information, assess the needs of a situation, examine the environment, and make propose explanatory ideas and theories to appease curiosity, understand the world, and make better decisions and predictions

The ventral visual pathway, and maybe other sensory systems, react to novelty (Bierderman 2006)

Defend Drive to defend ourselves and our valued accomplishments whenever we perceive them to be endangered

Amygdala (limbic region) (Carter 1998, p. 20)

The drive to acquire is a collection of some evolved psychological mechanisms to seek status, take control, retain objects and personal experiences that human’s value [13]. Humankind has been (and still is) driven to acquire goods that are either material, such as food, clothing, and shelter, or positional, such as social acknowledgement and recognition [17],[20]. The likelihood of survival was greater for those who were more apt at acquiring material goods, since doing so elevated their social status, made them (appear) more capable of caring and providing for others, and thus increased their chances of reproductive success. As a consequence, these individuals had to continue acquiring objects [20] because their social status and power were based on the continued well-being of their acquired dependents and goods [33].

The drive to bond is a categorization of evolved psychological mechanism to form social relationships and develop mutual caring commitments with other humans [13]. Our ancestors engaged in bonding activities to strengthen group cohesion on the inside and form coalitions

VOL. 1, NO. 2, Oct 2010 E-ISSN 2218-6301




83

against the outside. The premise is that those who bonded well had a relative advantage over those who did not. After all, establishing and maintaining groups of individuals bonded by mutual caring relationships improved the odds of surviving environmental threats [20]. Bonding and its associated aspects, such as trust, empathy, compassion, loyalty, respect, partnership, and alliance, also manifests itself in behavioral outcomes that include altruism and establishment of moral codes regarding social relationships [34],[35],[36],[37]. Interestingly, many of the strongest reactions, both positive and negative, are linked to belongingness and engagement in a mutually caring relationship [38]. The drive to comprehend is a clustering of evolved psychological mechanisms that push humans to collect information, assess the needs of a situation, examine their environment, and make observations about explanatory ideas and theories to appease curiosity and make sound judgments [13]. This mechanism encourages individuals to seek out information to resolve problems associated with fulfilling fundamental needs [24]. Individuals seek to learn in order to decrease their uncertainty, bring about closure to a problem that challenges well-being, appease curiosity that enhances well-being, or make situations more consistent with what is perceived as a “normal” behavior [24],[39],[40]. The drive to defend is a deep-rooted collection of evolved psychological mechanisms that make us defend ourselves and our valued accomplishments whenever we perceive them to be endangered. At the individual level, the drive to defend is activated by perceived threats to one’s person, valued objects, status, or beliefs [41]. At the collective or organizational level, the drive to defend triggers when individuals perceive a threat either to the bonds with others in their group or the collectively shared resources, or as a deviation from socially accepted norms deemed disloyal [41]. The human mind is preconditioned to enact to a variety of threats, and the reaction escalates as the severity of the threat heightens [42]. Lawrence and Nohria (2002) apply the Four-Drive model across individuals and groups to explain what motivates behavior in and by people in business organizations, the organization at large, and industries (the latter two as collective entities of humans possessing these innate drives that manifest in collectively to motivate behaviors). The drives are innate and ever-present, which remain dormant until spurred by environmental cues (e.g., the situation or circumstances presented) that has a moderating effect impacting motivation for display of a particular behavior such as decision-making or technology acceptance in this case. We propose that the environmental

cues conceptualized as the situation or circumstance presented to the person could be the task and the technology used for the task. The drives only manifest as motivating behavior when presented with the need to perform a task with a technology. A depiction of the model appears in Figure 1.

Figure 1: Four-Drive Model

The model has been applied in two empirical studies to explain how the four drives influence motivation of behavior, more specifically in terms of employee engagement, satisfaction, commitment, and intention to quit. The studies provide insights on what actions managers can take to satisfy the four drives to promote intended behaviors of their employees [10]. One study surveyed 385 employees from two global businesses, a financial services firm and a leading IS services business [10]. The other surveyed employees from 300 Fortune 500 companies [10]. Both studies indicated that “the ability of an organization to meet the four fundamental drives explains, on average, explained about 60 percent of employees’ variance on motivational indicators” [10] where previous models have only explained about 30 percent [10]. The study argues that “individuals and social institutions will enjoy adaptive advantage (i.e., advantages in meeting changing business environmental demands) to the extent that they are able to fulfill all four basic human drives” [13]. Providing appropriate tools, technologies, or opportunities appeases these drives and promotes desired behaviors of organizational workers [13]. Thus, the four-drive model and the underlying theory of

VOL. 1, NO. 2, Oct 2010 E-ISSN 2218-6301




84

evolutionary psychology provide a lens for enhanced understanding of human behaviors.

3. MANIFESTATION OF FOUR DRIVES IN TECHNOLOGICAL INNOVATIONS

We now provide examples of how each drive has manifested itself in relationship with a specific technological innovation, that is, how the drive has either amplified or dampened the adoption of an information system or technology in respect of what theories might have predicted (Table 2).

Table 2: Examples of the manifestation of the four drives

in IS and IT

Amplification of the Drive to Acquire Most open source software projects initially relied on programmers providing their services without receiving any of the economic rewards associated with market-based or firm models [43]. One explanation for the emergence of this phenomenon is that programmers participated in order to build a positive image within a peer group. Open source communities have reputation-based cultures [44],[45] and internal reputation markets [46]. The successful emergence of open source, despite the lack of economic gains, can be potentially explained by the opportunity it created to channel the drive to acquire, status and reputation of some software developers.

Dampening the Drive to Acquire Electronic brainstorming would seem to offer much to improve group decision-making [47], but the evidence is that it is not as effective as verbal brainstorming [48]. One explanation is that anonymity and the electronic recording of ideas divorce contributors from their ideas. There is no opportunity for participants to acquire or enhance status within the group [48]. Hence, for many, the drive to acquire

dampens the willingness to participate and thus lessens the likelihood of the adoption of electronic brainstorming.

Amplification of the Drive to Bond Short text messaging (SMS) was never designed for consumer use. It was a simple add-on with little forethought about design and potential use [49]. Typing messages on a small keyboard is cumbersome and limited to 160 characters. Nevertheless, SMS has been a great success. In 2005, five years after its introduction in Finland, nearly 1 billion text messages were sent by Finns [50]. A possible explanation for the success of SMS is that it supports the desire to bond through building, maintaining, and invigorating social relations [51]. SMS can be a form of gift-giving that mediates social relations [53]. Dampening the Drive to Bond Drive Amplify Dampen

Acquire Open source Anonymous brain storming

Bond SMS Telecommuting

Comprehend Web Hypercard

Defend Digital computing Health care

Telecommuting [52], for over a quarter of a century, has offered the prospect of reduced energy and real estate costs and improved job satisfaction. There is, however, little evidence of increased job satisfaction [53]. Many companies have policies to support telecommuting, but there are only a few telecommuters, which resulted in the verbiage of the so-called “telecommuting paradox” [54],[55]. One explanation is that telecommuting isolates workers both socially and professionally [56]. The inability to satisfy the drive to bond with fellow workers is a possible reason of this dissatisfaction. Thus, telecommuting research, which has had limited success in explaining what happens when people telecommute [53], could benefit from including the drive to bond as an explanatory variable.

Amplification of the Drive to Comprehend The HTML and HTTP, the foundations of the Web, emerged from CERN, the world’s largest particle physics laboratory. The Web was designed for sharing information among scientists. It was never intended to be a global information system to support a wide range of human activities (e.g., communication, commerce, entertainment, mass collaboration). HTML, created in 1990, has some serious shortcomings. It is a non-extensible language for describing the presentation of data that does not work well for large documents. XML, developed in 1996, in contrast, is an extensible language, focuses on the recording of data, and separates presentation and data. It was introduced to address the weaknesses of HTML [57]. Similarly, HTTP has some significant weaknesses: it is stateless and lacks security. Cookies, Java, and S-HTTP are among the technologies introduced to overcome HTTP’s deficiencies [58]. So, why did the Web have such a transformative impact on many

VOL. 1, NO. 2, Oct 2010 E-ISSN 2218-6301




85

fields beyond the physicists at CERN? It serves our drive to comprehend. We want to learn, our curiosity drives us to “google” in order to learn more about people we have met, to learn more about the news stories we have heard or the medical problems we might encounter, and so on. It does not matter to most people that the Web has a poor foundation. We put up with spam and fraud, because the drive to comprehend is so well served by the Web.

Dampening the Drive to Comprehend HyperCard, released in the 1987, preceded the Web and was highly popular in the Apple community. Applications, including an inventory management system for a global automotive manufacturer, were built by linking “cards into stacks.” The designer, Mark Atkinson, later recognized that he missed the mark. “I grew up in a box-centric culture at Apple. If I'd grown up in a network-centric culture, like Sun, HyperCard might have been the first Web browser. My blind spot at Apple prevented me from making HyperCard the first Web browser” [59]. His blind spot meant that HyperCard applications, and there were many in the education area [59], were shared physically in an ad hoc way. In addition, they ran only on Macintosh computers. Thus HyperCard stacks served inadequately the drive to comprehend. As he noted, “I thought everyone connected was a pipe dream," he said. "Boy, was I wrong. I missed that one” [59]. Connection is the key to serving the drive to comprehend, whether it is through the printing press, radio, television, or the Internet. A failure to connect, no matter how good the software, inhibits satiating the drive to comprehend.

Amplification of the Drive to Defend The history of digital computing is wedded to the drive to defend. From the early 1940s and until the early 1960s, the US armed forces funded the development of digital computing [60] in the interests of national security [61]. In 1951, General MacArthur told the U.S. Senate, “What I advocate is defend every place” [62], and massive sums were spent by the U.S. Department of Defense to meet this challenge. During the Cold War, integrated weapons and computing systems were deployed, and these could function seamlessly across the globe within the short period of a full-scale nuclear war. In sum, the most important legacy of the Cold War was digital computing [60]. Other key foundations of information systems, such as the Internet, started as ARPANET in 1969 by the Department of Defense to ensure secure communications during a war [63], were also propelled by the drive to defend.

Dampening the Drive to Defend The drive to acquire leads people to seek power and autonomy, and when these are threatened, the drive to defend kicks in to prevent loss of these valued attributes. The dampening effect of the drive to defend can partially explain the failure of some health care information systems [64],[65],[66]. Physicians have high status and autonomy, and strongly resist efforts to reduce either of these. The same resistance to the redistribution of power as the result of implementing an IS has been observed in other contexts (e.g., [67]).

A Caveat We are not claiming that the four drives are the only determinants of behavior in the situations we described in the preceding section, but we believe that they can contribute to our understanding of human behavior in an IS setting. The four drives provide additional key explanatory factors, and they help us to understand more precisely human behavior in a variety of situations, including multiple aspects of information systems (e.g., funding, adoption, use, and spread).

4. THE SCHOLARLY DILEMMA AND RESEARCH OPPORTUNITIES

We have presented the case for the inclusion of evolutionary psychology as a reference theory for IS from two perspectives. We first gave the general case for evolutionary psychology on the basis of manifold evolved psychological mechanisms that emerged to solve the problems of survival and reproduction faced by our distant ancestors. We then presented the Four-Drive model as a high level clustering of many of these modules, in much the same way that factor analysis reduces the complexity of interpreting a large number of variables. The dilemma is that we have two ways of thinking about the role of evolutionary psychology in IS research: we either think along evolved psychological mechanisms or along the four drives. Our solution is to have two modes in which scholars can deploy these models and associated research opportunities. We propose a new direction centered on understanding fundamental evolved psychological mechanisms and their impact on IS as well as an augmented current avenue that incorporates the four drives. A New Path of Inquiry for IS Our field has borrowed extensively; we have built upon and extended the theories of other fields. This might be our eternal fate because behavior and information are a

VOL. 1, NO. 2, Oct 2010 E-ISSN 2218-6301




86

successful evolutionary mix and as such no field can separate behavior from information. Those species with better information processing skills were more able to solve survival and reproduction problems and behaved accordingly. Nevertheless, we think we should not relinquish the pursuit of indigenous IS theory. The current dominant logic of the social science model of IS research is unlikely to lead to such theory because the borrowing paradigm is solidified within the editorial process. We need, in our opinion, to consider new research directions. First, we can work with neuroscientists and imaging technology to understand how a particular currently known module might influence human information processing. We could study how humans react to specific issues within the IS research domain. We might, for example, gain a better understanding of interface design and ease of use by developing an in-depth understanding of which modules handle the various elements of an interface and how they are handled. Second, and as part of this first path with the help of neuroscientists, we could seek to discover aspects of modules, or maybe even new modules, that have particular relevance to IS. The challenge is that all modules process information because of the very nature of the brain, so that we would need to focus on issues that might have particular pertinence to IS. For example, how do modules handle the perennial management challenge of differentiating between efficiency and effectiveness [68]? The brain needs efficiency in processing environmental signals so that threats are detected with sufficient time to react. It also requires effectiveness to make the right decision as a result of processing the received information. How are these issues resolved within and maybe across modules? What can they tell us for enhancing information systems? The potential of this new path is that it might lead us to uncover the deep structure of information systems and human information processing and a fundamental theory of IS. We will only know if some of us are willing to explore this new horizon and if our leading journals are sufficiently flexible to embrace a new paradigm and encourage it as an emerging trend. Augmenting the Social Science Path The four drive model fits easily into the social science paradigm. We can envision one or more drives as antecedents or moderators in behavioral IS models. Some studies thus far have considered the existence of human predispositions, mostly in form of traits, or IS outcomes [6],[7],[69],[70],[71],[72]. In addition, existing constructs that primarily base their definitional roots on social and

cognitive psychology may gain a richer conceptual understanding from integrating the four drives. We can also further develop an instrument to measure the drives, which has been done in other work (e.g., Nohria et al., 2008). The Four-Drive model essentially operationalizes evolved psychological mechanisms and enables us to add evolutionary psychology concepts to existing nomological IS networks (e.g., Junglas et al, 2009). Teaching Implications The introduction of evolutionary psychology continues to be met with heated opposition [73], slowing down the migration of some of the central ideas to other fields. As evidence accumulated, evolutionary psychology has become a more accepted lens to use in explaining human behavior [15]. As a result, it has gained academic ground in the psychology discipline [20]. Due to the original opposition, many scholars in unrelated fields are not sensitized to thinking about evolved psychological mechanisms, which is likely to be the case in the IS discipline. As far as teaching implications, Emerson noted, “people only see what they are prepared to see” [74] and the IS PhD curriculum rarely prepares IS scholars to consider the relevance of evolutionary psychology. Thus, if it is not part of an individual’s theoretical repertoire, which is typically formed by IS doctoral curricula, along with other lenses used for understanding human behavior, it is unlikely that one will see its possible explanatory power and miss opportunities to explain behaviors. Consequently, we urge those teaching doctoral seminars to include some readings on evolutionary psychology. Practical Implications Regarding practical implications, Nohria et al. (2008) exposed the value of four drives to aid managers in more precisely designing reward systems and facilitating conditions that appease these drives, which they assert are rooted in evolutionary psychology as innate motivators of behavior, which in turn motivated employees better than any extrinsically-oriented motivational techniques excluding the innate drives. This logic has practical implications for IS developers when designing systems that will enable the appeasement of the drives. Thus, during analysis, developers need to undertake a drives analysis to determine the potential impact of the four drives on implementation success. Will some stakeholders act defensively because of a loss of status? Will other stakeholders embrace the new system because it increases their ability to bond? We recommend that those who teach systems analysis and design make their

VOL. 1, NO. 2, Oct 2010 E-ISSN 2218-6301




87

students familiar with evolutionary psychology and the four drives model and their applicability to IS issues.

5. CONCLUSION We must remember that the person to whom we give today’s latest information system has an operating system founded on a set of evolved psychological mechanisms that have changed little since the beginning of our existence as a distinct species [75]. Since the human mind is the ultimate legacy system, it is prudent to explore the influence of those ancient mechanisms that direct how the brain formulates our behavior. There is considerable evidence that when psychologists took into account how the mind has evolved over time, their understanding of human behavior expanded, and this is also likely to be the case for understanding human reactions to information systems. In our opinion, evolved psychological mechanisms are important for understanding multiple dimensions of information systems and offer several opportunities for a deeper understanding of the information systems discipline on a theoretical and practical plane.

REFERENCES [1]Heisenberg, W. (1975). Bemerkungen über die

Entstehung der Unbestimmtheitsrelation. Physikalische Blätter, 31, pp. 193-196. (W.C. Price, S. Chissik, & W. Heisenberg, Trans. (1977)).

[2]Dimoka, A. (2010). What does the brain tell us about trust and distrust? Evidence from a functional neuroimaging study. MIS Quarterly, 34(2), 373-396.

[3]Dimoka, A., Pavlou, P., & F. D. Davis. (2007). Neuro-IS: The Potential of Cognitive Neuroscience for Information Systems Research - A Research Agenda. (Working Paper).

[4]Hantula, D. A., Brockman, D. D., & Smith, C. L. (2008). Online Shopping as Foraging: The Effects of Increasing Delays on Purchasing and Patch Residence. IEEE Transactions on Professional Communication, 51(2), 147-154.

[5].Hubona, G. S., and Shirah, G. W. (2006). The Paleolithic Stone Age Effect? Gender Differences Performing Specific Computer-Generated Spatial Tasks. International Journal of Technology and Human Interaction, 2(2), 24-46.

[6]. Junglas, I., Abraham, C., & Ives, B. (2009). Mobile Technology at the Frontlines of Patient Care: Understanding Fit and Human Drives in Utilization Decisions and Performance. Decision Support Systems, 46(3), 634-647.

[7]Kock, N. (2004). The Psychobiological Model: Towards a New Theory of Computer-Mediated Communication Based on Darwinian Evolution. Organization Science, 15(3), 327-348.

[8]Kock, N. (2009). Information systems theorizing based on Evolutionary Psychology: An interdisciplinary review and theory integration framework. MIS Quarterly, 33(2), 395-412.

[9]Dewsbury, D. (2009). Charles Darwin and psychology at the bicentennial and sesquicentennial: An introduction. American Psychologist, 64(2), 67-74.

[10]Nohria, N., B. Groysberg, & Lee, L. (2008). Employee motivation: A powerful new model. Harvard Business Review, 86(7), 78-84.

[11]Buss, D. M. (1999a). Evolutionary psychology: A new paradigm for psychological science. In D. H. Rosen & M. C. Luebbert (Eds.), Evolution of the psyche (pp.1-33). Westport, CT: Praeger.

[12]Buss, D. M. (1999b). Evolutionary psychology: The new science of the mind. Boston: Allyn and Bacon.

[13]Lawrence, P.R., & Nohria, N. (2002). Driven: how Human nature shapes our choices. San Franciso: Jossey-Bass.

[14]Wilson, E. O. (1975). Sociobiology : the new synthesis. Cambridge, MA.: Belknap Press of Harvard University Press.

[15]Alcock, J. (2001). The triumph of sociobiology. Oxford, New York: Oxford University Press.

[16]Cosmides, L. and Tooby, J. (2000). Evolutionary Psychology and the Emotions. In M. Lewis & J.M. Haviland-Jones (Eds.), The Handbook of Emotions, 2nd edition. New York: Guilford Press.

[17]Pinker, S. (1997). How the mind works. New York: Norton.

[18]Buss, D. (1995). Evolutionary psychology: A new paradigm for psychological science. Psychological Inquiry, 6, 1-30.

[19]Buss, D.M. (1996). The evolutionary psychology of human social strategies. In E.T. Higgins & A.W. Kruglanski (Eds.), Social Psychology: Handbook of Basic Principles, (pp.3-38). New York: Guilford Press.

[20]Cosmides, L., Tooby, J., & Barkow, J. (1992). Introduction: Evolutionary psychology and conceptual integration. In J. Barkow, L. Cosmides & J. Tooby (Eds.), The adapted mind (pp.3-15). New York: Oxford University Press.

[21]Hass, R.G., Chaudhary, N., Kleyman, E., Nussbaum, A., Pulizzi, A., & Tison, J. (2000). The Relationship Between the Theory of Evolution and the Social Sciences, Particularly Psychology. In D. LeCroy & P.

VOL. 1, NO. 2, Oct 2010 E-ISSN 2218-6301




88

Moller (Eds.), Evolutionary Perspectives on Human Reproductive Behavior (pp.1-20). New York: New York Academy of Sciences.

[22]Cosmides, L., & Tooby, J. (1994). Better than rational: evolutionary psychology and the invisible hand. American Economic Review, 84(2), 327-332.

[23]Tooby, J., & Cosmides, L. (2007). Evolutionary psychology, ecological rationality, and the unification of the behavioral sciences. Behavioral & Brain Sciences, 30(1), 42-43.

[24]Kaplan, S. (1992). Environmental preference in a knowledge-seeking, knowledge-using organism. In J. Barkow, L. Cosmides & J. Tooby (Eds.), The adapted mind (pp. 581-600). New York: Oxford University Press.

[25]Simons, R. C. (1996). Boo! - Culture, experience, and the startle reflex. New York: Oxford University Press.

[26]Ruth, W. (1993). Evolutionary Psychology and Rational-Emotive Theory: Time to Open the Floodgates. Journal of Rational-Emotive & Cognitive-Behavior Therapy, 11(4), 235-246.

[27]Moore, K. (2006). Biology as Technology: A Social Constructionist Framework for an Evolutionary Psychology. Review of General Psychology, 10(4), 285-301.

[28]Harré, R. (1998). The Singular Self: An Introduction to Psychology of Personhood. London: Sage.

[29]Kenrick, D., Schaller, M., & Simpson, J. (2006). Evolution is the new cognition. In M. Schaller, J. Simpson & D. Kenrick (Eds.), Frontiers of Social Psychology: Evolution and Social Psychology (pp.1-14). New York: Psychology Press.

[30]Sperber, D. (2005). Massive modularity and the first principle of relevance. In P. Carruthers, S. Laurence & S. Stich (Eds.), The Innate Mind: structure and contents. Oxford University Press.

[31]Tooby, J., & Cosmides, L. (1992). Psychological foundations of culture. In J. H. Barkow, L. Cosmides & J. Tooby (Eds.), The Adapted mind: Evolutionary psychology and the generation of culture (pp.19-136). New York: Oxford University Press.

[32]Lawrence, P. R. (2007). Being Human: A Darwinian Theory of Human Behavior. Online book retrieved from http://www.prlawrence.com/beinghumandownload.html

[33]Wilson, E. (2000). Sociobiology: The New Synthesis. Boston: Harvard University Press.

[34]Rusbult, C.E. & Van Lange. (2003). Interdependence, interaction, and relationships. Annual Review of Psychology, 54, 351-375.

[35]Trivers, R.L. (1971). The evolution of reciprocal altruism. Quarterly Review of Biology, 46, 35-57.

[36]Van Vugt, M., & Van Lange, P.A.M. (2006). The altruism puzzle: psychological adaptations for prosocial behavior." In M. Schaller, J. Simpson & D. Kenrick (Eds.), Evolution and social psychology (pp. 237-257). New York: Psychology Press.

[37]Wieselquist, J., Rusbult, C.E., Foster, G.A. & Agnew, C.R. (1999). Commitment, prorelationship behavior, and trust in close relationships. Journal of Personality and Social Psychology, 77, 942-966.

[38]Baumeister, R.F., & Leary, M.R. (1995). The need to belong: Desire for interpersonal attachments as a fundamental human motivation. Psychological Bulletin, 117, 497-529.

[39]Hackman, J.R., and Kaplan, R.E. (1974). Interventions into group processes: an approach to improving the effectiveness of groups. Decision Sciences, 5, 459-480.

[40]Kurzban, R., & Aktipis, C. A. (2006). Modular minds, multiple motives. In M. Schaller, J. Simpson & D. Kenrick (Eds.), Evolution and social psychology (pp.39-54). New York: Psychology Press.

[41]Hirschhorn, L. (1988). The Workplace Within: Psychodynamics of Organizational Life. Cambridge, MA: MIT Press.

[42]Buss, D. (2006). Evolution of aggression. In M. Schaller, J. Simpson & and D. Kenrick (Eds.), Evolution and social psychology (pp.263-286). New York: Psychology Press.

[43]Benkler, Yochai (2002) Coase's Penguin, or, Linux and the Nature of the Firm. The Yale Law Journal, 112(3).

[44]Feller (2000). (????) [45]Lerner (2002) (???) [46]Raymond (2001) (???) [47]DeSanctis, G., & Gallupe, R. B. (1985). Group decision

support systems: A new frontier. Data Base, 16(2), 3-10.

[48]Dennis, A.R., & Reinicke, B.A. (2004). Beta Versus VHS and the Acceptance of Electronic Brainstorming Technology. MIS Quarterly, 28(1), 1-20.

[49]Agar, J. (2003). Constant Touch: A Global History of the Mobile Phone. Cambridge, UK: Cambridge University Press.

[50]Kasesniemi, E.L., & Rautiainen, P. (2002). Perpetual Contact: Mobile Communication, Private Talk,

http://www.prlawrence.com/beinghumandownload.html

http://www.prlawrence.com/beinghumandownload.html

VOL. 1, NO. 2, Oct 2010 E-ISSN 2218-6301




89

Public Performance. New York: Cambridge University Press.

[51]Taylor, A.S., & Harper, R. (2003). The Gift of the Gab?: A Design Oriented Sociology of Young People's Use of Mobiles. Computer Supported Cooperative Work, 12(3), 267-396.

[52]Nilles, J.M. (1975). Telecommunications and Organizational Decentralization. IEEE Transactions on Communications, 23, 1142-1147.

[53]Bailey, D.E., & Kurland, N.B. (2002). A Review of Telework Research: Findings, New Directions, and Lessons for the Study of Modern Work. Journal of Organizational Behavior, 23, 383-400.

[54]Khalifa, M., & Davison, R. (2000). Viewpoint: Exploring the Telecommuting Paradox. Communications of the ACM, 43(3), 29-31.

[55]Westfall, R.D. (1997). The Telecommuting Paradox. Information Systems Management, 14(4), 15-20.

[56]Gainey, T.W., Kelley, D.E., & A., H.J. (1999). Telecommuting’s Impact on Corporate Culture and Individual Workers: Examining the Effect of Employee Isolation. SAM Advanced Management Journal, Autumn, 4-10.

[57]Watson, R.T. (2006). Data Management: Databases and Organizations. Hoboken, NJ: J. Wiley.

[58]Nikolik, D. (2003). A Manager's Primer on E-Networking: An Introduction to Enterprise Networking. Springer.

[59]Kahney, L. (2002). Hypercard Forgotten, but Not Gone. Wired. Retrieved from http://www.wired.com/gadgets/mac/commentary/cultofmac/2002/08/54365

[60]Edwards, P.N. (1997). The Closed World: Computers and the Politics of Discourse in Cold War America. MIT Press.

[61]Pugh, E.W., & Aspray, W. (1996). Creating the Computer Industry. Annals of the History of Computing, IEEE, 18(2), 7-17.

[62]US Senate, The Committee on Armed Services and the Committee on Foreign relations of the United States Senate. (1951). Conduct an Inquiry into the Military Situation in the Far East and the Facts Surrounding the Relief of General the Army Douglas Macarthur from His Assignments in the Area. Washington, DC.

[63]Ruthfield, S. (1995). The Internet's History and Development: From Wartime Tool to Fish-Cam. Crossroads, 2(1), 2-4.

[64]Beard, J.W., Keck, B., & Peterson, T. (2005). Information Systems and Health Care Vii When Success Results in Failure: The Challenge of Extending the It Infrastructure to Support Organ Procurement and Transplantation. Communications of AIS, 16(26), 517-538.

[65]Bhattacherjee, A., & Hikmet, N. (2007). Physicians' resistance toward healthcare information technology: A theoretical model and empirical test. European Journal of Information Systems, 16(6), 725-737.

[66]Lapointe, L., & Rivard, S. (2005). A Multilevel Model of Resistance to Information Technology Implementation. MIS Quarterly, 29(3), 461-491.

[67]Markus, M.L. (1983). Power, Politics, and Mis Implementation. Communications of the ACM, 26(6), pp. 430-444.

[68]Drucker, P. F. (2006). What executives should remember. Harvard Business Review, 84(2), 144-153.

[69]Agarwal, R. & Karahana, E. (2000). Time Flies When You're Having Fun: Cognitive Absorption and Beliefs about Information Technology Usage. MIS Quarterly, 24(4), 665-694.

[70]Kock, N. (2005). Media Richness or Media Naturalness? The Evolution of Our Biological Communication Apparatus and its Influence on Our Behavior Toward e-Communication Tools. IEEE Transactions on Professional Communication, 48(2), 117-130.

[71]Kock, N., & DeLuca, D. (2007). Improving Business Processes Electronically: an Action Research Study in New Zealand and the U.S. Journal of Global Information Technology Management, 10(3), 6-27.

[72]Kock, N., Chatelain-Jardón, R., & Carmona, J. (2008). An Experimental Study of Simulated Web-based Threats and Their Impact on Knowledge Communication Effectiveness. IEEE Transactions on Professional Communication, 51(2), 183-197.

[73]Segerstrêale, U. C. O. (2000). Defenders of the truth: The battle for science in the sociobiology debate and beyond. Oxford, New York: Oxford University Press.

[74]Emerson, R. W., & Emerson, E. W. (1904). The complete works of Ralph Waldo Emerson. Boston: Houghton Mifflin.

[75]Nicholson, N. (1998). How hardwired is human behavior? Harvard Business Review, 76(4), 134-147.

VOL. 1, NO. 2, Oct 2010 E‐ISSN 2218‐6301

Journal of Emerging Trends in Computing and Information Sciences ©2009‐2010 CIS Journal. All rights reserved.


90

Fruit Recognition using Color and Texture Features

S.Arivazhagan1, R.Newlin Shebiah1, S.Selva Nidhyanandhan1, L.Ganesan2

1Department of Electronics and Communication Engineering, Mepco Schlenk Engineering College, Sivakasi - 626 005, Tamilnadu, India.

2Department of Computer Science and Engineering, Alagappa Chettiar College of Engineering and Technology, Karaikudi -623 004, Tamilnadu, India.


ABSTRACT

The computer vision strategies used to recognize a fruit rely on four basic features which characterize the object: intensity, color, shape and texture. This paper proposes an efficient fusion of color and texture features for fruit recognition. The recognition is done by the minimum distance classifier based upon the statistical and co-occurrence features derived from the Wavelet transformed sub- bands. Experimental results on a database of about 2635 fruits from 15 different classes confirm the effectiveness of the proposed approach.

Key words: Fruit Recognition, Texture, Wavelet Transform, Co-occurrence Features.

1 INTRODUCTION

The computer vision strategies used to recognize a fruit rely on four basic features which characterize the object: intensity, color, shape and texture. This paper proposes an efficient fusion of color and texture features for fruit recognition. The recognition is done by the minimum distance classifier based upon the statistical and co-occurrence features derived from the Wavelet transformed sub- bands.

Recognition system has emerged as a ‘grand challenge' for computer vision, with the longer term aim of being able to achieve near human levels of recognition for tens of thousands of categories under a wide variety of conditions. The fruit recognition system can be applied for educational purpose to enhanced learning, especially for small kids and Down syndrome patients, of fruits pattern recognition based on the fruit recognition result [1]. It can be used in grocery store which makes the customers label their purchases using automatic fruit recognition based on computer vision. A number of challenges had to be overcome to enable the system to perform automatic recognition of the kind of fruit or vegetable using the images from the camera. Many kind of fruits are subject to significant variation in color and texture, depending on how ripe they are. For example, Bananas range from being uniformly green, to yellow, to patchy and brown.

Color and texture are the fundamental character of natural images, and plays an important role in visual perception. Color has been a great help in identifying objects for many years. It is often useful to simplify a monochrome problem by improving contrast or separation. The process of color classification involves extraction of useful information concerning the spectral properties of object surfaces and discovering the best match from a set of known descriptions or class models to implement the recognition task [2]. Texture is one of

the most active topics in machine intelligence and pattern analysis since the 1950s which tries to discriminate different patterns of images by extracting the dependency of intensity between pixels and their neighboring pixels [3] or by obtaining the variance of intensity across pixels [4]. Recently, different features of color and texture are combined together for their applications in the food industry [5].

Color features have been extensively applied for apple quality evaluation mostly for defect detection. For instance, color features of each pixel in images obtained in three components of RGB spaces could be successfully used to segment defects on ‘Jonagold’ apples [6, 7]. Tomato is another food product in which color features are widely used, as color is an indicator of the maturity of tomatoes. The early application of color features in tomato quality evaluation was preliminarily carried out by Sarkar and Wolfe [8] who used grey intensities of images to classify green and red tomatoes. Texture features are found to contain useful information for quality evaluation of fruit and vegetables, e.g., classification of grade of apples after dehydration with the accuracy of 95% [9], and prediction of sugar content of oranges with a correlation coefficient of 0.83 [10].

Recently, different features of color, size, shape, and texture are combined together for their applications in the food industry. Normally, by increasing the features used, the performance of the methods proposed can be increased. Moreover, both surface information (color and texture) and geometry information (size and shape) of food products in images play a significant part in defect detection and class discrimination [11]. Thereby, to capture more proper information about the quality of food products from images, multiple kinds of features corresponding to the grading system of the food products should be proposed.

Color and texture features are used to locate green and red apples [12]. Here, the texture property


VOL. 1, NO. 2, Oct 2010 E‐ISSN 2218‐6301



91

plays two roles in the recognition procedure. Texture based edge detection has been combined with redness measures, and area thresholding followed by circle fitting, to determine the location of apples in the image plane. It was shown that redness works for red apples as well as green apples. This increased texture contrast helped to identify apples separately from background. Three features analysis methods color-based, shape based and size-based are combined together in order to increase accuracy of recognition [1].

An unified approach that can combine many features and classifiers, where all features are simply concatenated and fed independently to each classification algorithm. The fusion approach is validated using a multi-class fruit-and-vegetable categorization task in a semi-controlled environment, such as a distribution center or the supermarket cashier. The results show that the solution is able to reduce the classification error in up to 15 percentage points with respect to the baseline [13].

The paper is structured as follows. The next section discusses the Proposed Method. The Section 3 gives the Recognition Results and Discussion. Finally, Section 4 gives the Concluding remarks of the proposed method.

2. METHODOLOGY

The two sections that involved in this work are Training and Classification. The block diagram of the proposed method is given in Figure 1.

Figure 1: Fruit Recognition System

The proposed Fruit recognition system, shown in Figure 1, need a change in the color space of the images, in order to obtain one channel containing the luminance information and two other channels containing chrominance information. The HSV representation is often selected for its invariant properties. The hue is invariant under the orientation of an object with respect to the illumination and camera direction and hence more suited for object retrieval. Texture features are computed from the luminance channel ‘V’, and color features are computed from the chrominance channels ‘H’ and ‘S’ [14]. The component which corresponds to brightness of the color (V) is decomposed using Discrete Wavelet Transform [15-17]

and the co-occurrence matrix is constructed from the approximation sub-band by estimating the pair wise statistics of pixel intensity. The use of the co-occurrence matrix [18,19] is based on the hypotheses that the same grey-level configuration is repeated in a texture. Further, co-occurrence features such as contrast, energy, local homogeneity, cluster shade and cluster prominence are calculated from co-occurrence matrix C(i,j), derived for transformed sub-bands and stored in the features library. There exist 5 co-occurrence features i.e., texture features for an image. Statistical features such as Mean, Standard Deviation, Skewness and Kurtosis are derived from H and S components. Hence there will be 8 chrominance or color statistical features for an image. Thus a total of 13 features characterize one fruit image. The feature extraction process is illustrated in Figure 2.

Figure 2: Feature Extraction

In the classification phase, for the test fruit image, color and texture features are derived as that of the training phase and compared with corresponding feature values, stored in the feature library. The classification is done using the Minimum Distance Criterion. The image from the training set which has the minimum distance when compared with the test image says that the test image belongs to the category of that training image.

3. RESULTS AND DISCUSSION Data Set

The Supermarket Produce data set comprising 15 different categories: Plum, Agata Potato, Asterix Potato, Cashew, Onion , Orange , Taiti Lime , Kiwi , Fuji Apple, Granny-Smith Apple , Watermelon, Honeydew Melon, Nectarine, Williams Pear and Diamond Peach ; totalizing 2633 images are used for experimental purpose. Fig. 3 depicts some of the classes of the data set. These fruit images are divided into training and testing set, where 50% of the fruit images from each group are used to train the system and the remaining images serves as the testing set. The number of images used for training and classification for each type of fruits is shown in Table I.

VOL. 1, NO. 2, Oct 2010 E‐ISSN 2218‐6301



92

Figure 3: Dataset used for Fruit Recognition

All of the images were stored in RGB color-

space at 8 bits per channel. The images were gathered at various times of the day and in different days for the same category. These features increase the data set variability and represent a more realistic scenario. Fig. 4 shows an example of Kiwi and Granny-Smith Apple categories with different lighting. The differences are due to illumination, no image pre-processing was performed. The Supermarket Produce data set also comprises differences in pose and in the number of elements within an image. Fig. 5 shows examples of the Cashew category. Note that there are variations in the pose of the Cashew’s plastic repository. In addition, Fig. 6 shows the variability in the number of elements within an image.

Figure 4: Illumination Differences With in Categories

Figure 5: Pose Differences: Cashew

Figure 6: Variability on Number of elements: Plum

Table 1: List of Images in Database

No. of Fruit

Images used for S.No. Fruits

Total No. of Fruit Images

Training Testing

1 Agata Apple 201 100 101 2 Asterix

Apple 182 91 91

3 Cashew 210 105 105 4 Diamond

Peach 211 105 106

5 Fuji Apple 212 106 106 6 Granny-

Smith Apple 155 77 78

7 Honeydew Melon

147 73 79

8 Kiwi 171 85 86 9 Nectarine 247 123 124 10 Onion 75 38 37 11 Orange 103 51 52 12 Plum 264 132 132 13 Spanish Pear 159 79 80 14 Taiti Lime 106 53 53 15 Watermelon 192 96 96

Total 2635 1314 1326 Preprocessing

For a real application in a supermarket, it might be necessary to cope with illumination variations, sensor capturing artifacts, specular reflections, background clutter, shading, and shadows. Therefore, in order to reduce the scene complexity, it might be interesting to perform background subtraction and focus in the object’s description.

Background subtraction is a commonly used class of techniques for segmenting out objects of interest in a scene. The name “background subtraction” comes from the simple technique of subtracting the observed image from the estimated image and thresholding the result to generate the objects of interest.

The best channel to perform the background subtraction is the S channel of HSV-stored images. This is understandable, given that the S channel is much less sensitive to lighting variations than any of the RGB color channels.

Algorithm for Extracting Region of Interest

1. The input fruit image is converted to HSV colour space.

VOL. 1, NO. 2, Oct 2010 E‐ISSN 2218‐6301



93

2. Perform thresholding operation on the S component, since S is much less sensitive to lighting variation. 3. Close small holes using the Closing morphological operator with a disk structuring element. 4. Find the area of the Region of Interest from the binary image. 5. Crop the Region of Interest and replace the binary values with the original pixel intensity.

Figure 7: Extracting Region of Interest from the Image

The role of color descriptors has been demonstrated to be quite remarkable in many visual inspection tasks. In some other tasks, texture measurements are needed because of unevenly colored or achromatic surfaces. In many applications, color and texture must be combined to achieve good performance. At the same time, the computational complexity of the methods must be kept as low as possible.

The fruit images are converted to HSV color space and the Statistical features such as Mean, Standard Deviation, Skewness and Kurtosis are derived from H and S components. Hence there will be 8 color statistical features for an image. The V component is subjected to one level decomposition using Discrete Wavelet Transform and the co-occurrence features such as contrast, energy, local homogeneity, cluster shade and cluster prominence are derived from the Co-occurrence

matrix constructed from the approximation sub-band. There exist 5 co-occurrence features for an image. First, the fruit recognition system is evaluated with color and texture features individually. With the set of 8 statistical features the recognition rate obtained for the individual fruit images is shown in Table II.

Table 2: Results of Fruit Recognition System

Recognition Rate

S.No. Fruits Using

Colour Features

Using Texture Features

Using Colour

and Texture Features

1 Agata Apple

56.435 74.257 95.049

2 Asterix Apple

52.747 65.934 90.109

3 Cashew 77.1428 94.2800 99.047 4 Diamond

Peach 45.283 55.660 75.471

5 Fuji Apple 34.9056 78.3018 82.073 6 Granny-

Smith Apple

30.769 89.743 96.153

7 Honeydew Melon

66.216 76.056 95.945

8 Kiwi 32.558 47.6744 58.139 9 Nectarine 32.258 74.1935 79.032 10 Onion 43.24 78.378 86.486 11 Orange 30.769 40.384 69.230 12 Plum 48.484 84.090 89.393 13 Spanish

Pear 32.500 60.000 86.25

14 Taiti Lime 58.490 88.679 98.1132 15 Watermelon 40.625 55.208 89.583 Total 45.49483 70.85591 86.00488

Hence, the color and texture information are complementary and when used together they yield good results of classification. 4. CONCLUSION

The use of computers to analyze images has many potential applications for automated agricultural tasks. But, the variability of the agricultural objects makes it very difficult to adapt the existing industrial algorithms to the agricultural domain. The proposed method can process, analyze and recognize fruits based on color and texture features. In order to improve the functionality and flexibility of the recognition system shape and size features can be combined together with color and texture features. Further, by increasing the number of images in the database the recognition rate

VOL. 1, NO. 2, Oct 2010 E‐ISSN 2218‐6301



94

can be increased. This algorithm can be used for smart self service scales. ACKNOWLEDGMENT The authors are grateful to acknowledge the financial support of Department of Science and Technology, New Delhi for carrying out this Research project. The authors are expressing their sincere thanks to the Management and Principal of Mepco Schlenk Engineering College, Sivakasi for their constant encouragement and support. REFERENCES [1] Woo Chaw Seng and Seyed Hadi Mirisaee, “A New Method for Fruits Recognition System”, MNCC Transactions on ICT, Vol. 1, No. 1, June 2009. [2] Ferat Sahin, “A Radial Basis Function Approach to a Color Image Classification Problem in a Real Time Industrial Application”, Master's thesis, Virginia polytechnic institute, Blacksburg, 1997. [3] Kartikeyan,B and Sarkar,A, “An identification approach for 2-D autoregressive models in describing textures” Graphical Models and Image Processing, vol.53, pp.121-131, 1991. [4] Haralick, R. M., Shanmugan, K. and Dinstein, I., “Textural features for image classification”, IEEE Transactions on Systems, Man, and Cybernetics, vol.3, pp.610-621,1973. [5] Jain, A and Healey,G, “A multiscale representation including opponent color features for texture recognition”, IEEE Transactions on Image Processing vol.7, No.1, pp. 124-128, 1998. [6] Leemans, V. and Destain, M.-F, “A real-time grading method of apple based on features extracted from defects” Journal of Food Engineering, vol.61, pp.83-89,2004. [7] Leemans, V., Magein, H. and Destain M.-F, “Defects segmentation on ‘Golden Delicious’ apples by using colour machine vision” Computers and Electronics in Agriculture, vol.20, pp.117-130,1999. [8] Sarkar, N, and Wolfe, R. R, “Feature extraction techniques for sorting tomatoes by computer vision” Transactions of the ASAE, vol.28, pp.970-979, 1985.

[9] Fernández, L., Castillero, C. and Aguilera, J. M., “An application of image analysis to dehydration of apple discs” Journal of Food Engineering, vol.67, pp.185-193, 2005. [10] Kondo, N., Ahmad, U., Monta, M. and Murasc, H., “Machine vision based quality evaluation of Iyokan orange fruit using neural networks”, Computers and Electronics in Agriculture, vol.29,pp.135-147, 2000. [11] Paliwal, J., Visen, N. S., Jayas, D. S. and White, N. D. G., “Cereal grain and dockage identification using machine vision” Biosystems Engineering, Vol.85, pp. 51-57, 2003. [12] Zhao, J.T., J. Katupitiya, J., “On-tree fruit recognition using texture properties and color data”, International conference on Intelligent Robots and Systems, pp. 263-268, 2005. [13] Anderson Rocha, Daniel C. Hauagge, Jacques Wainer, Siome Goldenstein, “Automatic fruit and vegetable classification from images”, Computers and Electronics in Agriculture, Vol. 70, pp. 96–104, 2010. [14] Drimbarean, A., Whelan, P. F., “Experiments in colour texture analysis”, Pattern Recognition Letters, Vol.22, No.10, pp.1161-1167,2001. [15] Daubechies, “Orthonormal bases for compactly supported Wavelets”, Comm. Pure Applied Math., Vol. XLI, NO. 41, pp. 909-996, 1988. [16] Strang. G and Nguyen.T, “ Wavelets and Filter Banks”, Wellesley-Cambridge Press, Wellesley, MA,1996. [17] Burrus,Goinath.R.A and Guo.H, “ Introduction to Wavelets and Wavelet Transform – A Primer”, Prentice Hall,1998, Expansion of notes for the IEEE Society’s tutorial program held in conjunction with the ICASSP,1993. [18] Argenti, F., Alparone, L., and Benelli, G., “Fast algorithms for texture analysis using co-occurrence matrices”, IEE Proceedings-F, Vol. 137, No. 6, pp. 443-448,1990. [19] Haralick .M, “Statistical and structural approaches to texture”, Proceedings of the IEEE, Vol.67, No. 5, pp786-804, 1979.




Solving Fuzzy based Job Shop Scheduling Problems using Ga and Aco Surekha P1, S.Sumathi2

1 Research Scholar, EEE, PSG College of Technology, Coimbatore

2 Asst. Professor, EEE, PSG College of Technology, Coimbatore


ABSTRACT

In this paper, we present a genetic algorithm and ant colony optimization algorithm for solving the Job-shop Scheduling Problem (JSSP). The genetic algorithm comprises of different stages like generating the initial population, selecting the individuals for reproduction and reproducing new individuals. Ant Colony Optimization (ACO) is a metaheuristic inspired by the foraging behavior of ants, which is also used to solve this combinatorial optimization problem. In JSSP ants move from one machine (nest) to another machine (food source) depending upon the job flow, thereby optimizing the sequence of jobs. The sequence of jobs is scheduled using Fuzzy logic and optimizes using GA and ACO. The makespan, completion time, makespan efficiency, algorithmic efficiency and the elapsed time for the genetic algorithm and the ant colony algorithm are evaluated and compared. The improvement in the performance of the algorithms based on the computed parameters is also discussed in this paper. Computational results of these optimization algorithms are compared by analyzing the JSSP benchmark instances, FT10 and the ABZ10 problems. Keywords : Fuzzy Logic, Planning, Scheduling, Makespan, Genetic Algorithm, Ant Colony Optimization.

1. INTRODUCTION Job Shop Scheduling Problem (JSSP) is one of the most difficult combinatorial optimization problem, which is used in complex equipment manufacturing system to validate the performance of heuristic algorithms. The research on JSSP not only promotes the development of relative algorithms in the field of artificial intelligence, but also provides means of solutions and applications for complex JSSP. JSSP can be thought of as the allocation of resources over a specified time to perform a predetermined collection of tasks. Job shop scheduling has received a large amount of attention, because it has the potential to dramatically decrease costs and increase throughput, thereby, increasing the profits in automation industries. Job-shop is a system that processes n number of tasks on m number of machines. The total ordering defines a set of precedence constraints, meaning that no activity can begin execution until the activity that immediately precedes it in the complete ordering has finished execution. Each of the m activities in a single job requires exclusive use of the resources defined in the problem [4]. No activities that require the same resource can overlap in their execution and once an activity is started it must be executed for its entire duration. Usually, these orders differ in terms of processing requirements, materials needed, processing time, processing sequence and setup times. Job-shop problems are widely known as a NP-hard (Non-Polynomial Deterministic) problem and commonly defined as a set of jobs whose operations are to be processed in an uninterrupted manner on a given machine for a specified length of time. GA is a local search algorithm that follows the evolution paradigm. Starting from an initial population, the algorithm applies genetic operators in order to produce offsprings, which are presumably more fit than their ancestors [7]. Every individual in the population

represents a solution at the end of every generation. The major strength of GA when compared with other local search algorithms lies in the fact that in a GA framework more strategies can be adopted together to find individuals to add to the mating pool, both in the initial population phase and in the dynamic generation phase. Such an adaptation allows the search space to be explored at every algorithm step. The Ant Colony System (ACS) algorithm is a distributed algorithm which is extensively used to solve NP-hard combinatorial optimization problems. Its original model is based on the foraging behavior of real ants who find an approximately shortest way to the food by detecting the density of pheromone deposited on the route [5]. In real ants, the term pheromone denotes the chemical substance deposited by ants as they move, but in artificial ants it acts like an attraction for the other ants to follow. In general, ants create pheromone paths from their nests to the available food sources and the shortest path is the one with highest pheromone concentration between the source and the destination. In JSSP, the time required for all operations to complete their processes is called makespan. In this paper, the objective is to minimize this makespan value by applying the genetic algorithm and the ACO algorithm. When the makespan is minimized, at least one of the optimal solutions is active so that no operation can be processed without violating the technological constraints. For this reason, every time when makespan is optimized, a schedule can be described by the processing orders of operations on the machines. Giffler and Thompson [9] proposed algorithms using priority dispatching heuristics to solve production scheduling problems for single and multiple schedules. They demonstrated that every optimal schedule is the same as an active optimal schedule. Brooks and White

95




[10] developed a procedure for finding optimal solutions for the production scheduling problem incorporating a wide variety of possible criteria and constraints, using the algorithm developed by Giffler and Thompson. Blackstone [11] provides a comparison of several dispatching rules. Their work illustrated several measurements criteria that were used to evaluate dispatching rules and identified several rules that exhibit good overall performance. Cheng [12] was the first to provide a tutorial survey of recent works on solving classical JSSP using Genetic Algorithms (GA) and addressed the key issue of encoding a GA for solving the JSSP. Carlos A. Coello Coello et al. [1] proposed an ant based system to solve the JSSP and experiments show that the proposed approach can reduce the number of evaluations performed without a degradation of performance. The performance evaluation of optimization techniques are based on computational intelligence paradigms for solving job shop scheduling problem to obtain optimized completion time, to decrease the makespan of job sequence and in turn to increase the feasibility of scheduling process. The organization of the paper is as follows: Section 2 gives a brief introduction about job shop scheduling problem, and Section 3 describes the planning and scheduling of jobs using fuzzy logic. Section 4 explains the algorithm of JSSP using GA and section 5 delineates the application algorithm of ACO for JSSP. The simulation results are analysed in section 6 and section 7 concludes this paper. 2. JOB SHOP SCHEDULING PROBLEM Recently, researchers have been focusing on investigating machine scheduling problems in manufacturing and service environments where jobs represent activities and machines represent resources, and each machine can process one job at a time.

Fig 1 Stages in Solving JSSP

Fig 1 illustrates the stages involved in solving the job shop scheduling process in this paper. Planning of machines and scheduling of jobs to each machine with respect to levels is performed for better optimization results. Fuzzy logic is used to implement the planning and scheduling of jobs. The scheduled job sequences are then optimized using GA and ACO and the best algorithm is determined based on parameters like makespan, mean

completion time, mean waiting time, least completion time and algorithmic efficiency. Job-shop scheduling is an activity that comprises of a set of jobs to be processed on a set of machines. The job-shop scheduling problem can be defined as the allocation of machines over time to perform a collection of minimizing or maximizing a specific performance measure while satisfying the operation precedence constraints, machine capacity constraint, processing time and ready time requirements. The resource used in an industrial environment is known as machines and the basic task modules which performs the operation is called job [14]. Each job may be comprised of several elementary tasks called operations, which are interrelated by precedence constraints. The processing of an operation requires the use of a particular machine for an uninterrupted duration, called the processing time of the operation. Each machine can process only one operation at a time. The routing, processing times and precedence constraints are specified by a process plan. The main distinction between the classic flow-shop and a job-shop is that, in the former case each job passes the machines in the same order whereas in the latter case the machine order may vary per job. Since workflow in a job-shop is not unidirectional, scheduling becomes quite tedious. For a particular job-shop process plan, several feasible schedules can be generated and measured. The processing order on each machine that minimizes the corresponding cost is desired by the objectives such as minimization of process cost, makespan and flow time or maximization of throughput, systems/resource utilization and production rate. The job-shop scheduling problem (JSSP) can be described as a set of n jobs denoted by Jj where j =1, 2... n. It has to be processed on a set of m machines denoted by Mk where k =1, 2… m. Operation of j th job on the k th machine will be denoted by Ojk with the processing time p

jk. Once a machine starts to process a job, no interruption is allowed. Each job is composed of a sequence of operations and the operation order on the machines is pre-specified. Each operation is characterized by the required machine and the fixed processing time. The process plan specifies the routing, processing times and precedence constraints among operations of each job. There are several constraints on jobs and machines: • The same machine does not encounter more than one job • No priority is defined for the operations defined in the job • No operation can be preempted • A machine can process only one job at a time 3. PLANNING AND SCHEDULING The fuzzy parameters used for assigning jobs to machine with respect to their processing time are tabulated in Table 2. Here numbers of input parameters are jobs and levels and output parameter is processing time. Ten jobs and ten machines is taken as test samples and processed for planning and scheduling from the benchmark instances. Based on the processing time fuzzy logic is used to group the crisp values interms of linguistic variables

96




like short processing time (SPT), medium processing time (MPT) and long processing time (LPT). A large computation effort may be actually saved by avoiding failures with problems whose lack of feasibility is due to insignificant constraint violations [3]. The computation effort can be actually saved by performing a proper planning strategy. The procedure for planning using fuzzy logic is shown below: Step 1: Assign jobs and levels as the input variables Step 2: Assign the processing time as the output variable Step 3: Formulate the rules based on triangular membership function with the linguistic parameters such as SPT, MPT and LPT. Step 4: Sequence the jobs depending upon the rule base. Step 5: Repeat step 3 and 4 for each level. Step 6: Stop the planning phase

Table.1. Scheduled Sequences of Jobs a. Fisher and Thompson 10x10 instance

The scheduling of the jobs with respect to machine

and level is shown for both benchmark FT10 and ABZ10 in Table 1.a and 1.b. The job is aligned in the order for machines with the data available in the benchmark instances. The time taken by a job to wait in a queue is calculated as waiting time. The jobs aligned to each machines with respect to levels are tabulated to show the job sequence ordering according to the given benchmark instance. The job sequence after scheduling is optimized to rearrange the job sequence so that less waiting time and minimum makespan are obtained. Once the priority rules are evaluated and the jobs are scheduled then the system is prepared for optimizing the job sequence.

b. Adams, Balas, and Zawack 10x10 instance

4. GENETIC ALGORITHM FOR JSSP Job-shop scheduling is one of the most interesting and challenging applications for GA’s. The major difference among different genetic algorithm applications is the chromosome representation. A very important issue in building a GA for the job-shop problem is to devise an appropriate representation of solutions together with problem-specific genetic operations. This enables all the chromosomes to be generated, either in the initial phase or during the evolutionary process to produce feasible schedules [14]. GA have been used for a wide variety of problems such as machine learning, cellular manufacturing, and combinatorial optimization, inventory control, traveling salesman and game playing. The application of GA’s in scheduling was introduced by Davis and Liepins to investigate the simplest scheduling problem of a static queue of jobs with specified due dates and run times without precedence constraints. Gupta studied a single machine model with an objective to minimize flow time variance. Lee and Kim examine the performance of a parallel GA for a model in which earliness and tardiness penalties are allowed to be arbitrary and jobs share a common due date. Cheng consider a model of identical parallel machines where the objective is to minimize the maximum weighted absolute lateness about an unrestricted common due date [1]. The makespan of the operation is used as fitness value. Makespan is denoted as Cmax is the time when the last operation leaves the workplace which is given by the equation (1).

( )nCCCC ,...,2,1maxmax = Where,

(1) ∑−

+=n

kkjmjkj WC

1)( )()( ρ

Cj is the completion time of job j, Wjk is the waiting time of job j at sequence k and p jm(k ) is the processing time needed by job j on machine m at sequence k .The GA consists of four main stages: evaluation, selection,

97




crossover and mutation. The evaluation procedure measures the fitness of each individual solution in the population and assigns it a relative value based on the defining optimization (or search) criteria. The algorithm of GA used for optimizing Job Shop Scheduling problem is summarized as follows: Step 1: Initialization - The chromosomes consist of genes which define the assignment of operations to the machines. Every chromosome in the population represents a solution for the problem. Step2: Initial population - Based on the scheduling phase the initial population is formed considering the processing time and the operations assigned on the machines. The size of the population depends upon the number of jobs and levels in the benchmark instance. Step 3: Fitness Evaluation – For each chromosome the fitness function is evaluated according to Eq. (1), the objective to minimize the makespan. Step 4: Selection – The best chromosomes are chosen for reproduction using the method of tournament selection. Step 5: Genetic operators – Crossover and mutation are applied to the selected chromosomes and new individuals are generated until a maximum is reached. Step 6: Termination –If maximum number of generations is reached then stop else continue from Step 3. 5. ANT COLONY OPTIMIZATION BASED JSSP Ant colonies exhibit very interesting behaviours, though one specific ant has limited capabilities, the behaviour of a whole ant colony is highly structured. They are capable of finding the shortest path from their nest to a food source, without using visual cues but by exploiting pheromone information [5]. While walking, ants can deposit some pheromone on the path. The probability that the ants coming later choose the path is proportional to the amount of pheromone on the path, previously deposited by other ants. This theory was the basis for forming the Ant Colony Optimization (ACO) algorithm using artificial ants. The artificial ants are designed based on the behavior of real ants. They lay pheromone trails on the graph edges and choose their path with respect to probabilities that depend on pheromone trails and these pheromone trails decrease progressively by evaporation. At the end of each generation, each ant present in the population spawns a complete tour traversing all the nodes based on a probabilistic state transition rule. The nodes are chosen by the ants based on the order in which they appear in the permutation process. The node selection process involves a heuristic factor as well as a pheromone factor used

by the ants. The heuristic factor, denoted by ij, and

the pheromone factor, denoted by , are indicators of how good it seems to have node j at node i of the permutation. The heuristic value is generated by some problem dependent heuristics whereas the pheromone factor stems from former ants that have

found good solution. The next node is chosen by an ant according to the following rule that has been called pseudo random proportional action choice rule [6]. With probability q0, where 0≤ q0<I is a parameter of the algorithm, the ant chooses a node from the set of nodes (s) that have not been selected so for which maximizes , where α≥0 and β≥0 are constants that determine the relative influence of the pheromone values and the heuristic values on the decision of the ant. The probability of choosing the next node is chosen from the set S according to the probability distribution given by:

βα ητ )()( ijij

∑∈

=

Shijij

ijijijP βα

βα

ητητ

)()()()(

This probability also known as the transition probability is a trade-off between the pheromone factor and the heuristic factor. The heuristic factor is computed as

SjXF jij ∈= ,)(

1η , where F(X) represents the

cost function of X. While constructing its tour, an ant will modify the amount of pheromone on the passed edges by applying the local updating rule

0)()1()( ρττρτ +−← tt ijij , where )(tijτ is the amount of pheromone on the edge (i, j) at time t; ρ is a parameter governing pheromone decay such that 0 < ρ < 1; and 0τ is the initial value of pheromone on all edges. Once all ants have arrived at their destination, the amount of pheromone on the edge is modified again by applying the global updating rule

)()()1()( ttt ijijij τρτρτ ∆+−← , where

, if this is the best tour, otherwise 1)( −=∆ Ltijτ0)( =∆ tijτ , and L indicates the length of the globally

best tour. The pheromone updating rule was meant to simulate the change in the amount of pheromone due to both the addition of new pheromone deposited by ants on the visited edges and to pheromone evaporation [2]. The algorithm stops iterating either when an ant found a solution or when a maximum number of generations have been performed. The ACO procedure for the JSSP is shown below: Step 1: Initialization Let Mij denote the machine index Assign no. of ants = no. of jobs Each job is a set of operations denoted by {O11,O12,…, OJOBS,1} Initialise parameters α, β, ρ and φ Record the current position of ants Step 2: Building ant tour If ant k is the first operation then Assign the first operation Ok1 of job 1 to ant k’s first visited operation Record the visited operation of ant k in its memory Else

98




Choose the next operation Ok2 based on constraints of the job If next operation is not the last then Assign Mij after next operation Determine the transition probability based on

∑∈

=

Shijij

ijijijP βα

βα

ητητ

)()()()(

Step 3: Apply the local pheromone updating rule 0)()1()( ρττρτ +−← tt ijij , where )(tijτ is the amount of pheromone on the edge (i, j) at time t; ρ is a parameter governing pheromone decay such that 0 < ρ < 1; and 0τ is the initial value of pheromone on all edges. Step 4: Apply the global updating rule

)()()1()( ttt ijijij τρτρτ ∆+−← , where

, if this is the best tour, otherwise 1)( −=∆ Ltijτ0)( =∆ tijτ , and L indicates the length of the globally

best tour. Determine the time taken by the ant k and compute the shortest time Step 5: If stopping condition is true then stop else continue from Step 2. 6. EXPERIMENTAL RESULTS The Fisher and Thompson 10x10 instance (FT10) and Adams, Balas, and Zawack 10x10 instance (ABZ10) benchmark problems were used to compare and analyze the performance of computational intelligence paradigms for solving job shop scheduling problem. The procedures of GA and ACO were simulated using MATLAB R2008b on Intel core 2 Duo (1.73GHz), 1GB RAM PC. A. Scheduling The job-shop is concerned with the simultaneous and synchronized ordering of operations on several machines. Based on each level, sequence of the jobs will vary, and hence scheduling has to be processed before optimizing the jobs. For the FT10 and ABZ10 benchmark, during scheduling, the machines will process different jobs in different levels irrespective of the processing time. The Waiting Time (WT), Completion Time (CT) and its Priority Level (PL) of corresponding job are computed after the job passes through all machines so that all operations are completed. The waiting time, completion time are graphed for comparison and priority level corresponding to their completion time is ranked and plotted in graph which is shown in Fig 2 and Fig 3 respectively. For the FT10, the makespan value is calculated as 1356 sec and for ABZ10 the makespan is 1908 sec. Thus scheduling of the both benchmark instance are computed and the makespan value is obtained. This makespan value is kept as reference for optimization using computational intelligence techniques.

Fig 2 Graphical representation of CT and WT

Fig 3 Priority Level Vs Jobs for FT10 and ABZ10

B. Genetic Algorithm Genetic algorithms are applied to solve a problem using the principle of evolution. In the search process it will generate a new solution using genetic operator such as selection, crossover and mutation. The search procedure will stop once there is no improvement in next iteration. Here each chromosome is represented by a list of job order, and the machine sequence taken from the FT10 and ABZ10 benchmark problems. The algorithmic parameters used for simulating JSSP using GA are: Population size : based on level and machine (for most of the instances it was 10) Crossover probability : 0.6 Mutation probability : 0.0333 Crossover : Two point crossover Mutation : Bit flip Number of generations : 100 The Waiting Time (WT), Completion Time (CT) and the Priority Level (PL) of the corresponding job are evaluated during the GA generational run for several iterations. The graphical representation of the computed waiting time, completion time and the priority levels are shown in Figs 4 and 5 for both FT10 and ABZ10 problems.

99




Fig 4 Graphical representation of CT and WT for

FT10 using GA

Fig 5 Priority Level Vs Jobs for FT10 using GA

Fig 6 Graphical representation of CT and WT for

ABZ10 using GA The makespan value is computed from these evaluations as 1046 sec for the FT10 problem and it is found that the makespan value is decreased while optimizing the problem using GA when compared to the scheduling phase. Similarly the ABZ10 problem was also simulated and the results are shown in Figures 6 and 7 respectively. Using GA for the ABZ10 problem the makespan value is found to be 1769sec, which is a decreased value when compared with the makespan obtained in the scheduling phase.

Fig 7 Priority Level Vs Jobs for ABZ10 using GA

C. Ant Colony Optimization In JSSP, the ants construct a sequence by first choosing a job for the first position, then a job for the second position, and so on until all jobs are scheduled. The parameters of the ant system for solving the FT10 and ABZ10 problems are divided into two groups: those that influence the state transition (a and b) and those that determine the pheromone update (the evaporation constant r and the number of ants m). From simulation it appears that the parameters Q (pheromone allocation per unit of distance) and t0 (initial pheromone level) are of very little importance to the algorithm’s performance. The parameters and their values are used for running the ACO algorithm is shown below: Number of ants = based on level and machine Weight of pheromone trail α = 1 Weight of heuristic information β = 1 Pheromone evaporation parameter ρ = 0.7 Constant for pheromone updating Q = 10 The ACO algorithm for the JSSP problem was tested on both FT10 and ABZ10 problems and the makespan was computed as 1052 sec and 1502 sec respectively. The performance characteristics of the problems based on waiting time and completion time are shown in Fig 8. The priority level of the jobs is shown in Fig 9.

Fig 8 Graphical representation of CT and WT using

ACO

100




Fig 9 Priority Level Vs Jobs using ACO

D. Comparative Analysis The waiting time, completion time and the priority levels for FT10 and ABZ10 are tabulated for the Scheduling phase, GA and ACO in Table 2. The makespan value for the FT10 and ABZ10 problem has decreased for both GA as well as ACO when compared with the value obtained in the scheduling phase. For the FT10 problem, GA computed the makespan as 1046 sec and ACO as 1052 sec, which indicates that performance of the algorithm depends upon the problem defined. While considering the ABZ10 problem, it can be seen that ACO performs better compared to GA, since the makespan for ACO is 1502 sec and this is a reduced value than 1769 sec, the makespan of GA.

Table 2 Comparison of Benchmark Instances

In Table 3, the various parameters such as Makespan (MS), Mean Completion Time (MCT), Least Completion Time (LCT), Mean Waiting Time (MWT), Least Waiting Time (LWT), Elapse Time (ET), MS Reduction Efficiency and Algorithm Efficiency for the optimization techniques

with the corresponding benchmark is compared and analyzed. Applying ACO for FT10, the Mean Completion Time is reduced to 785.3 sec while the other techniques GA and the scheduling obtained higher MCT. For ABZ10, the mean completion time is reduced to 1101.7 sec using Ant Colony Optimization.

Table 3 Comparative Analysis of GA and ACO

The Least Completion Time is also less for ACO, 542 sec for FT10 and 728 sec for ABZ10. Comparing mean waiting time, the ant colony optimization performs better with least MWT for FT10 with 269.9 sec and for ABZ10 with 323 sec. The time taken for the algorithms to perform is known as the elapsed time. The elapsed time for the performance of ant colony optimization is more when compared with GA. The elapse time, MS reduction efficiency and algorithm efficiency for scheduling cannot be computed, since the makespan value for scheduling is manipulated manually. With the overall performance, it can be concluded that ACO is superior than genetic algorithm for solving JSSP.

7. CONCLUSION This paper presents a novel knowledge-based approach for the job shop scheduling problem (JSSP) by utilizing the various constituents of the computational intelligence techniques such as Genetic Algorithm (GA), and Ant Colony Optimization (ACO). The well known Fisher and Thompson 10x10 instance (FT10) and Adams, Balas, and Zawack 10x10 instance (ABZ10) problem is selected as the experimental benchmark problem and simulated using MATLAB R2008b. This research focused primarily on discovering new approaches that can match the computational intelligence techniques in solving Job Shop Scheduling problems. Significant improvements can be made by modifying the goals of this paper and adopting techniques to extend the knowledge of job shop scheduling problems. The research dealt specifically with the classical 10x10 job shop scheduling problem with the objective of minimizing the makespan. The research can

101




[7] Pezzella F, et al. [2007] "A genetic algorithm for the Flexible Job-shop Scheduling Problem." Computers and Operations.

be extended to a larger size problem to analyse the performances. Dynamic and stochastic job-shop problems which incorporates attributes such as non-zero ready times, inter – travel time, multiple scheduling time, multiple job routes simultaneously, assign weights to jobs and uncertain processing times can also be solved using the computational intelligence techniques. These features increase the search space of solutions and make the task of scheduling very complicated. This provides a wide area for research giving rise to potential areas in the scheduling problems to be solved.

[8] Fevrier Valdez, Patricia Melin ,Olivia Mendoza, [2008] “A new evolutionary method with fuzzy logic for combining particle swarm optimization and Genetic algorithm: the case of neural networks optimization”, IEEE, Vol. 7, pp 45-49. [9] Giffler, B. and Thompson,G.L. [1960] "Algorithms for solving production scheduling problems. Operations Research." Operations Research, vol. 8, pp. 487-503

[10] Brooks, G.H. and White,C.R. [1965] "An algorithm for finding optimal or near optimal solutions to the production scheduling problem." Journal of Industrial Engineering, vol. 1, pp. 34-40

REFERENCES [1] Coelle, Carlos A. [2007] "An Ant System with steps counter for the Job Shop Scheduling Problem." IEEE trans. Evolutionary Computation, pp. 477-484. [11] Blackstone Jr., J.H.,Philips,D.T., and Hogg,G.L.

[1982] "A state-of-the-art survey of dispatching rules for manufacturing job-shop operations." International Journal of Production Research, vol. 20, pp. 27-45

[2] Dorigo, Marco. [1997] "Ant Colony System: A Cooperative Learning Approach to the Traveling Salesman Problem." IEEE Transactions on Evolutionary Computation, vol 1. [12] Cheng, R.,Gen,M. and Tsujimura,Y. [1995] "A

tutorial survey of job-shop scheduling problems using genetic algorithms." Computers and Industrial Engineering, vol 37, pp. 51-55

[3] Dubois, Didier. [1993] "Fuzzy constraints in job-shop scheduling." IJCAI'93 Workshop on Knowledge-Based Scheduling.

[13] Xiaoyu Song ,Limei Sun ,Qiuhong Meng [2008] “Deadlocks solving strategies in hybrid PSO algorithm for job shop scheduling” IEEE, Vol. 4, pp71-75.

[4] J. Christopher Beck, Patrick Prosser and Evgeny Selensky. [2002] "Graph Transformations for the Vehicle Routing and Job Shop Scheduling Problems." In Graph Transformation, Germany: Springer-Verlag, pp. 60-74. [14] Innani, Alok D. [2004] ‘Applying Data Mining to Job

Shop Scheduling Problems using Regression Analysis’ Ohio University, Japan.

[5] Jun Zhang, Xiaomin Hu, X. Tan, J.H. Zhong and Q. Huang. [2006] "Implementation of an Ant Colony Optimization for job shop scheduling problem." Transactions of the Institute of Measurement and Control, 93-108.

[15] Jain.A.S, Meeran.S, [1999] “Deterministic Job-Shop Scheduling:Past,Present and Future”, European Journal of Operational Research,113(2), pp.390-434. [16] Li Xiaoping, [1999] “Solving Job Shop Scheduling Problem using Demarcation-Genetic algorithm”, Electric Machines and control,3(2), pp. 93-96.

[6] Merkle, Daniel. [2000] "An Ant Algorithm with a new pheromone evaluation rule for total tardiness problem." Real World Applications of Evolutionary Computation, Springer, Gremany. [17] http://www.swarmintelligence.org

[18] http://www.aco-metaheuristic.org [19] http://people.brunel.ac.uk/

102




A Systems Approach for Dealing with Resistance to Change: With Reference to Library and Information Professionals Working in Academic and

Research Sector Libraries in India

Kshema Prakash Deputy Librarian

Indian Institute of Technology Rajasthan, Camp Office: MBM Engineering College Jodhpur – 342011, Rajasthan, India Email: [email protected]

ABSTRACT

Library and information landscape is changing rapidly and continuously in response to the dynamic changes in the environment. This paper attempts to identify the causes of resistance among library and information professionals in India and suggests measures of improvement for a smooth transition in this scenario. The authors examine the concept of systems approach and its application to library and information field. A systems model for dealing with resistance among the library and information professionals for a positive acceptance of change and rendering better service is recommended. Keywords: Change Management, Resistance, Library and Information Professionals (LIPs), Systems Approach. 1. INTRODUCTION

Change is the order of the world. It is the change, particularly, developmental changes that have been key drivers in societal transformation at large. The present stage of the society is that of information based knowledge society. Libraries are in existence ever since the recorded knowledge has started to be preserved for future. The traditional concept of library is being redefined from a place to access paper records or books to one that also houses the most advanced media. Libraries are changing in response to changes in the learning and research environment and also changes in the expectations of library users. These changes are evolutionary. Consequently, Library and Information Professionals (LIPs) are increasingly combining traditional duties with tasks involving changing technology. Traditional library performance measures fail to explain fully what is happening in libraries today because their scope is too narrow to encompass the field of change. 2. LITERATURE REVIEW Nature of Change in Information Services: According to Lyndon Pugh [1] the nature of change in information services characterized by the factors like diversity and unpredictability of the services and staffing of libraries, they are cross-boundary specialization, the structural change in library and information services, the amount of complexity driven by mixed economy, personalization of library and information services, competition and/or collaboration with network giants like Google etc. All these constantly demand new skills and competencies over the traditional ones.

Fig. 1 Nature of change in information services (Source:

Pugh, L., 2007)

Characteristics of emerging library and information environment: Modern library and information environment is characterized by electronic communication, both synchronous and asynchronous, web-based information sources, multimedia information, and is uncontrolled largely as a result of the Internet facilitating information creation, distribution and access. Accordingly, typical user expectations of the present day include – everything in full text and downloadable or printable; faster service; uninterrupted service availability, virtual reference service librarian available online, easy access; easy-to-use web resources permitting self-service; a librarian who knows all subjects and all databases; everything should be in electronic or digital format; several options / alternatives to choose from; a library web site that is capable to conduct all library transactions online viz., library registration, document delivery request, loans and renewals, etc.; and a web search engine to find required information.

103





Dynamics of Library and Information Profession: The next few decades will continue to be ones of transition and role redefinition for the LIPs. Even as they grow and make an effort to stay current in the rapidly changing technology environment, they will always be faced with the challenge of simultaneous learning, implementation and planning. The LIPs can no longer afford to remain institutionalized passive spectators, instead they have to find new ways to add value and remain relevant in this rapidly changing confusing and competitive environment. All the activities will now have to be tailored to give long distance and often home delivered information, which is the demand of time. Following table illustrates some sources of challenges faced by the LIPs in the present environment.

Table 1: Sources of challenges faced or implications on Library and Information Professionals due to various

attributes, and their responses to them. (Adopted from: Cardina, C. and Wicks, D., 2004 [2])

Attribute Impact / Challenge

Response

Developments in ICTs

Self-sufficient users, less library use, reluctance to use physical materials

Innovation-driven and customer oriented, development of electronic and digital collections, providing proactive support and training to users.

Changing economy

Demand for speedy delivery of information and increasing e-commerce

Adopting consortial power to obtain better prices of information resources, self-sustenance through marketing of library services, and adapting business system design methods to library management.

Changing education and learning environment

Life-long learning emphasis, information divide – info rich and info poor due to monopoly of publishers and copyright, subject specialization diversity

Making the library fully accessible both physically and intellectually via electronic networks, and by providing ICT support for remote users like off-campus users or distance learners on a 24-hour basis, developing co-operative online learning and educational programmes, incorporating into the teaching and

learning process. Changes in scholarly communication

Increase in journal costs, shift from acquisition to access and licensing models, changing preservation methods, archiving, copyright and access restrictions

Advising users on publishing and journal cost-effectiveness, finding alternatives to print resources, increasing e-resource subscriptions, resource sharing and cooperative acquisitions, developing institutional repositories and encouraging open access publications and open archive initiatives.

The biggest challenge for Indian librarianship is to

bring about attitudinal change among the library staff. Libraries and librarians are still the lowest priority in the decision making process and the librarians are least visible persons. If this main challenge is confronted by librarians and libraries in real earnest it is only then that one can expect all the necessary changes within the system itself. But a serious pre-requisite to overcome this challenge is to upgrade the competencies and skills of the library and information professionals, since it is imperative to become deserving before asking for something. Ingraining these competencies will motivate the LIPs to serve their customers in a better manner, thus helping them to contribute towards reaching organizational goal. A major road block in this task often confronted by LIPs is the resistance to change. Resistance to Change: Change is a ubiquitous element of organizational life; so too is resistance to change. Understanding why and how opposition to change occurs, and developing the ability to respond effectively to manifestations of change resistance, is crucial to the success or failure of organizational change.

Change, in organizations as elsewhere, involves moving from a known state to a new state – one that is to some degree unknown. Change involves letting go of things as they are in order to take up new ways of doing things. Organizational change challenges the statusquo and it may challenge the values and perceived rights of workers and workgroups. People working in organizations respond to change in many different ways. For some, change is welcome – fresh and exciting. These people, the early adopters and change leaders, tend to welcome and embrace change enthusiastically. Others may be more cautious – responding to change by seeking to test and examine changes before proceeding. For some, change may threaten their established values and understandings and therefore be deeply unsettling. For this latter group change is something to be strongly resisted.

104




The action of opposing something that we disapprove or disagree with can be called as ‘resistance’. Introduction to change is a highly complex process. The uncertainties caused by the expected change and in-equilibrium, as a consequence of changes, sometimes results in resistance to change. The resistance may adversely affect the positive features of an individual or organization and make the individual or the organization counter-productive. Resistance to change arises from individual’s attitude rather than technical issues of change. 1) Causes of Resistance to Change in Library and Information Centres: Some of the causes of resistance to the changes occurring in the environment and resistance towards acceptance of these changes in the present library and information scenario can be explained as follows:

i) Fear of change: It is recognized that change, above all technological change, often produces fear and anxiety in Library and Information Professionals. Internet related technologies are pushing them towards the establishments of new information spaces such as digital libraries, technology centres, and learning resource centres. Here it becomes necessary to involve staff with varying levels of knowledge and expertise. However, there are many difficulties in letting people accept change. Some will be optimistic and proactive, while others will feel threatened and react with resistance.

ii) ICT as a sense of lost control: The Internet has also represented for some professionals a sort of attack on the prestige of their career. With the technologies readily available at every one’s desk, some librarians began to feel left on the margins, because patrons really seemed to make it work without their help. It was like a direct assault to the role of gatekeepers which librarians had assumed believing they knew better than those whom they serve.

iii) Unsure scenarios: Unpredictable future regarding the role of library and information professionals that creates a sense of uncertainty among professionals.

iv) Technostress: Another important issue related to technology when we deal with networking, is technostress. Today librarians have to shift their focus from relatively stable and knowable local collections to a plethora of information sources with diverse characteristics located anywhere on the global information network. A variety of new tasks, skills and competencies are required by those who in general understand the benefits that applications of modern technologies are meant to achieve. Technostress affects staff and users as libraries offer more and more information through websites and other remotely accessible electronic systems.

v) Lack of standardization: Several authors have lamented over the years the lack of (classification, cataloguing, retrieval) standards when their working life comes to terms with the

Internet and electronic resources. Libraries today are increasingly acquiring and providing access to electronic resources of any kind. This results in increasing expenditures of money and workload for librarians without any certainty about the preservation of these resources.

vi) Lack of quality: A major concern among librarians seemed to be that of the poor quality of information delivered by Internet resources which are not controlled or short of organization. There is no overall structure on the Internet that allows reference librarians to navigate to quality information claiming for the definition of standards for the description and classification of Internet resources.

vii) Competition of search engines and commercial tools etc.: Finally it seems that since the advent and development of the World Wide Web (WWW), users have been migrating to commercial services and search engines to fulfill their information needs, tools that are thus seen as concurrents to reference librarians’ work. WWW appears to be the major information provider to patrons, librarians should work towards offering viable alternatives to their commercial competitors [3].

3. THE RESEARCH STUDY Methodology: A study of library and information professionals was conducted by administering a structured questionnaire. The objectives are to find causes of resistance among the LIPs towards change and to suggest measures for a smooth transition.

• Profile of respondents: The respondents for the survey were Library and Information Professionals (LIPs) of academic institutions (universities) and research organizations. They are serving in executive/managerial cadre in these information centres.

• Data Characteristics: One hundred (100) copies of questionnaire were administered to randomly selected LIPs and valid responses were received from 88 LIPs across India after pursuasion. a. Library wise distribution: The libraries were

categorized into two groups viz., academic and research libraries. Out of 88 valid responses, 42 (47.73%) were from academic libraries and 46 (52.27%) were from research libraries.

b. Age wise distribution: The age of respondents was categorized into two classes. 20-40 years category is for young professionals, whereas 40-60 is for senior professionals. 37.50% (33) respondents are in the age group of 25-40 and 63.5% (55) respondents are in the age group of 41-60 years.

c. Qualification wise distribution: The qualifications are classified into four levels as follows.

105




i. B.L.I.S –Bachelors in Library and information science with a master’s degree in any other subject.

ii. M.L.I.S – Masters in Library and information Science with or without any other master’s degree.

iii. M.Phil – Master of philosophy in library and information science.

iv. Ph.D. – Doctorate in Library and information Science.

v. Others According to the results, 1.14 % of respondents carry a bachelor’s degree in Library and Information Science with a master’s degree in some other subject. 55.68% of respondents possess a master’s degree in Library and Information Science with or without another maters degree. 5.68% respondents are Master of Philosophy in Library and information science. Whereas 35.23% of respondents hold doctorate in Library and Information Science. Those having other qualifications constitute to 2.27% of the sample.

d. Gender wise distribution: Out of 88 respondents 18 were women and 70 were men. That corresponds to 20.45% of female respondents and 79.55% male respondents.

Primary Data Findings: The causes of resistance faced by the library information professionals have been identified based on the systems approach as discussed earlier. Accordingly ten statements relating to the causes of resistance faced by them were framed. The statements are:

R1. Lack of proper attitude towards change causes resistance.

R2. Poorly motivated personnel pose resistance to change.

R3. Inadequate training causes resistance. R4. Difficulty in understanding fast and complex

changes in the environment sets off resistance. R5. Deep rooted dogmas and technophobia produce

resistance towards technology coupled change. R6. Underdeveloped competencies cause resistance

for the transformed environment. R7. Unavailability of adequate resources and

infrastructural facilities triggers resistance. R8. Lack of customer orientation creates resistance. R9. Absence of management support causes

resistance. R10. Resistance is raised due to absence of

supportive / conducive work atmosphere. The responses to these statements elicited by the

library and information professionals in the questionnaire were used to collect information regarding resistance posed by the LIPs and are summarized below. For convenience, the statements are coded as R1, R2, R3….R10.

Responses are categorized on a five point Likert scale from Strongly Agree (1) to Strongly Disagree (5). Likert scaling is a bipolar scaling method, measuring either

positive or negative response to a statement. This technique presents a set of attitude statements. Subjects are asked to express agreement or disagreement of a five-point scale. Each degree of agreement is given a numerical value from one to five. Thus a total numerical value can be calculated from all the responses. Since these statements are also related to attitude towards change i.e., resistance, Likert scaling technique was used. R1: Lack of proper attitude of LIPs towards change Many times when the employees fail to comprehend changes happening around and in their profession it is observed that they do not display proper attitude towards change. They may not participate actively, rather choose to be passive. This may result in resistance. From the 79 (89.76%) responses generated, for weights 1 and 2, indicate that the professionals agree that lack of proper attitude towards change results in resistance. R2: Poor motivation Motivation is both internal as well as external factor. Also it can be positive and negative. When there is very low motivation either internally or externally an employee may not understand completely the need for change and may resist. Even when there is some negative motivation or discouragement, resistance may creep in. From the 73 (82.96%) responses generated, for weights 1 and 2, indicate that the respondents agree that poor motivation in the employees causes resistance, while 7 (8%) of them cannot determine whether this is a cause and 8 (9%) of them do not agree to the statement. R3: Inadequate training to support change Change that has been planned by the authorities, if implemented without fulfilling the prerequisites like providing the employees with necessary training to accept and support the desired change, such a situation leaves ample scope for resisting change in employees. From the 71 (80.68%) responses generated for weights 1 and 2, it is indicated that the professionals feel inadequate training leads to resistance. While 8 (9%) of them could not decide whether this can be cause, 9 (10%) of them do not agree to this fact. R4: Difficulty in understanding fast and complex changes The pace of changes taking place in the external and internal environments in which libraries operate is very rapid. Even before a new technology can be completely understood and implemented it becomes obesolete due to proliferation of another competent technology. This creates a lag in understanding, appreciating and accepting the change. The 62 (70.46%) responses generated for weights 1 and 2, indicate that the library and information professionals agree that difficulty in understanding the fast and complex changes taking place in the information environment produce resistance. Out of 88 respondents 10 (11%) of them opted to be neutral, whereas, 16 (18.18%) of them did not agree to this reason. R5: Deep rooted dogmas and technophobia This statement relates to the set traditions, procedures, methods etc. that have been followed by librarians since ages. These traditional methods of managing libraries may not match with the current requirements if the library has

106




to remain relevant in the society. Similarly these deep rooted doctrines and dogmas may have an impact on the employee to make him a technophobe. These may one of the probable reasons to pose resistance. The 56 (63.63%) responses generated for weights 1 and 2, indicate that the respondents agree that deep rooted dogmas and technophobia could form a probable cause for resistance among the professionals to embrace change. While 15 (17%) of them have chosen to remain non-committal to any opinion, 17 (19%) out of 88 respondents do not agree with this view. R6: Underdeveloped competencies Improved education and training are necessary to assist individuals in developing the competencies they will need on the job. The competencies required by a library and information professional have been discussed in the previous sections. The concept of core competencies involves knowledge, which should be acquired through formal education and training; skills or abilities, which are acquired through practice; and attitudes, which involve emotional and social aspects. Underdevelopment of such essential competencies may lead to lack of comprehension of change and its subsequent acceptance. This may inturn lead to resistance [4]. The 65 (73.86%) responses generated for the weights 1 and 2, indicate that underdevelopment of competencies can lead to resistance in library and information professional, towards change. Also, 8 (9%) of 88 respondents did not comment anything upon it and 15 (16.91%) of them chose to disagree to this statement. R7: Inavailability of adequate resources Acceptance of change is a process involving various steps. Although the employees are competent enough, they are trained and motivated to an optimal level to accept change to implement technologically advanced quality services in libraries, if adequate resources are not provided to implement the desired services, this may prove fatal. Such a situation will lead to considerable resistance to change. Out of 88 (100%), 65 (73.86%) respondents think that non availability of adequate resources and infrastructure will lead to resistance, while 8 (9%) cannot opine anything regarding this statement. 15 (16.91%) of respondents do not agree to this factor being a cause of resistance to change amongst library and information professionals. R8: Lack of customer orientation When a library and information professional is unaware or does not seek to understand his customer / users’ requirement, he will not be in a position to extend service as per the users’ needs. When he is not willing to alter the way in which he provides service, he fails to customize service according to the users’ needs. This is a form of resistance to change and customization of service. 69 (78.4%) out of 88 (100%) respondents feel that lack of customer orientation is a cause for resistance towards change, while 5 (6%) of them cannot say anything regarding this. But 14 (16.36%) of them have opposed this view. R9: Absence of management support Inspite of the professional’s willingnes to provide new or customized services to the users’, he may sometimes not

receive support from his management in terms of identification, acknowledgement and appreciation, he may not continue to provide the same. In turn he may pose resistance to any further change initiatives. As many as 76 (86.36%) respondents agree that absence of management support will cause resistance among the library and information professionals. While 5 (6%) of them do not lament anything upon this statement, 7 (7.95%) of them show disagreement. R10: Lack of conducive work atmosphere Work atmosphere plays a vital role in productivity of employees. It also influences the attitude of employees. Improper work atmosphere may cause stress and burnout in library and information professionals. That too in wake of so many changes taking place rapidly, lack of conducive work atmosphere can overstress them. As a result they may not be interested in facing any further changes, thus posing resistance even to positive or productive changes. As many as 71 (80.68%) respondents agree that if the work environment is not conducive to the professionals, they will mount resistance. 5 (6%) of them do not say anything about it, whereas, 12 (13.63%) show disagreement to this statement. 4. RESULTS & DISCUSSION Data Analysis: Data analysis is conducted in three phases

as follows: A. Semantic Differential Profile B. Factor Analysis - to identify the significant

combination of causes of resistance to change surfaced by the professionals.

C1. Comparison of mean scores of responses of academic sector and research sector to test the difference of opinions between groups, which is further validated by z-test.

C2. Comparison of mean scores of responses of junior and senior age groups corresponding to 20-40 yrs. and 40-60 yrs. respectively, further validated by z-test.

The data collected from the responses are analyzed by using suitable statistical techniques such as Mean Comparison, t-test and z-test using MS Excel and SPSS packages. Mean score comparison was carried out based on type of libraries and age. A majority of respondents were male members and there was a skewness towards the qualification of master’s degree and above among the respondents. Due to this skewed distribution of profiles of respondents, mean score comparison based on these two categories was not done. A. Semantic Differential Profile: Semantic differential

profile is a graphical representation of the mean weights of the responses given by the respondents related to causes of resistance in the questionnaire. The semantic differential profile of the probable causes of resistance to change, surfaced by the library and information professionals indicates that there is general agreement regarding the statements related to causes of resistance. They agree strongly on the following aspects.

107




• Lack of proper attitude towards change. • Inadequate training. • Non-availability of adequate resources and

infrastructural facilities. • Absence of management support.

B. Factor Analysis: Factor analysis was deployed to find

out the significant factors that cause mounting of resistance from the library and information professionals towards change. Using data from the large sample, factor analysis applies an advanced form of correlation analysis to the responses to a number of statements. The purpose of this analysis is to determine if the responses to the statements are highly correlated. The results of the factor analysis are as follows.

Table 2: Total Variance Explained Initial Eigen Values Rotation Sums of

Squared Loadings Co

mponent Total % of

Variance

Cumulative %

Total % of Varian

ce

Cumulative

% 1 3.912 39.123 39.123 2.443 24.434 24.434 2 1.398 13.980 53.103 2.080 20.803 45.237 3 .989 9.888 62.991 1.775 17.755 62.991 4 .865 8.655 71.646 5 .657 6.568 78.214 6 .638 6.381 84.595 7 .589 5.887 90.482 8 .439 4.386 94.868 9 .319 3.188 98.056 10 .194 1.944 100.000

Three factors were extracted using principal component analysis and varimax rotation.

Table 3: Factor Loading of Causes of Resistance

Component 1 2 3 R1- Attitude .682 -.151 .127 R2 – Motivation .719 .195 -.066 R3 – Training .255 .234 .539 R4 –Fast and complex changes .687 .271 .313

R5 - Technophobia .677 .366 .342 R6 – Competencies .582 .123 .561 R7 – Resources and infrastructure .040 .106 .889 R8 – Customer orientation .314 .692 -.020 R9 – Management support -.101 .792 .341 R10 – Work atmosphere .135 .790 .170

Fig. 2: Semantic Differential Profile for responses related to resistance posed by library and information professionals

Factor 1 is Personal Attributes. It has a positive loading with Lack of proper attitude, Poor motivation, Difficulty in understanding fast and complex changes in the environment, Deep rooted dogmas and technophobia, and Underdeveloped competencies. This explains that the respondents have a consensus regarding the facts that lack of proper attitude towards change, difficulty in understanding fast and complex changes taking place in the information environment cause building barriers or resistance to change. They also agree that poorly motivated personnel, with ingrained technophobia due to deep rooted dogmas pose resistance towards technology coupled change initiatives. There is a common agreement regarding underdeveloped competencies being a resisting factor. Factor 2 is Work Environment. It has a positive loading with Lack of customer orientation, Absence of management support, and Absence of conducive work atmosphere. The consensus of respondents over this factor indicates that support from higher authorities or management, and creation and maintenance of a conducive or suitable work atmosphere plays a major role in managing resistance. Lest, absence of management support and conducive work atmosphere will become sources of resistance among library and information professionals. Also they agree that lack of customer or user oriented thinking serves in building up resistance. Factor 3 relates to Support Facilities. This shows a positive loading with Inadequate training, Underdeveloped competencies, and non-availability of adequate resources and infrastructural facilities. The analysis indicates that management support and provision of adequate infrastructural facilities by management plays a major role in bringing about change. Absence or lack of these components will contribute to mounting up resistance. The professionals also feel that inadequate training to develop skills to face challenges brought in by changes, will also contribute towards building up resistance in library and information centres. The three factors have Eigen values equal to or

greater that 1.0 indicating that they best fit the data obtained from the responses to the challenges in part 2 of the questionnaire. Also, the three factors together explain almost 62.99% of the total variance in the responses to the statements. C1. Comparison of mean scores – Academic vs.

Research: The mean scores of the responses elicited to the statements relating to challenges in the part 2 of questionnaire are tabulated below in two categories viz. Academic Library Professionals and Research

108




Library Professionals followed by its graphical representation for a clear picture.

Table 4: Mean Score of Academic and Research Sector

LIPs Library Type R1 R2 R3 R4 R5 R6 R7 R8 R9 R10 Academic Library 1.83 2.05 2 2.09 2.29 2.19 1.67 2.19 1.76 1.98 Research Library 1.83 1.78 2 2.33 2.37 2.24 2 2.24 1.87 2.11

Fig. 3: Comparison of mean scores of responses elicited by LIPs

of academic and research sectors The above graph indicates that inspite of majority

consensus between both the groups towards the general nature of the causes of resistance to change, there is little difference in their opinions in certain areas like poor motivation in professionals, difficulty in understanding fast and complex changes, non-availability of adequate resources and infrastructural facilities.

In order to explain the difference, a two-tailed z-test for each attribute was further carried out with 0.05 level of significance. A z-test is a statistical test used in inferencing if the difference between two population means is significant based on the difference between two sample means. According to the Central Limit Theorem, the sampling distribution of difference between means approach that of a normal distribution as the samples are large. The table below gives the difference between the Z scores.

Table 5: Z test for Resistance related statements

Zc = +1.96

α = 0.05

A-MS

(x1) R-MS

(x2) A-SD σΑ R-SD σR Zs

Value

Sig / NSi

g

R1 1.82926

8 1.82608

7 0.91930

5 0.67673

6 0.017 NS

R2 2.04761

9 1.78260

9 1.01097

3 0.75757 1.402 NS

R3 2 2 0.98711

2 0.89442

7 0 NS

R4 2.09523

8 2.32608

7 1.05482

7 1.01224

4 -1.054 NS

R5 2.28571

4 2.36956

5 1.13235

6 1.01890

3 -0.36 NS

R6 2.19047

6 2.23913 0.99359

2 1.03676

4 -0.23 NS

R7 1.66666 2 0.72133 0.84327 -2.07 S

7 6 4

R8 2.19047

6 2.23913 1.08735

7 1.01510

3 -0.18 NS

R9 1.76190

5 1.86956

5 0.84995 0.88465

2 -0.6 NS R10 1.97619

2.108696 0.84995

1.037695 -0.66 NS

R1 to R10 – Attributes related to causes of resistance A-MS (x1)– Mean Scores of Academic Sector Responses R-MS (x2) – Mean Scores of Research Sector Responses A-SD (σA) – Standard Deviation of Mean Scores of Academic Sector Responses

R-SD (σR) - Standard Deviation of Mean Scores of Research Sector Responses ZC – Z Critical Value ZS – Z Static Value (Calculated) α Level of significance of difference in variance Sig / NSig – Significant / Not Significant µ - The mean of the population

Fig. 4 Sampling distribution of difference between means If the difference between the observed sample means

falls within the acceptance region it inferences that there is no difference between population means scores of responses. There were no significant differences observed between the mean scores of two groups for each of the ten attributes relating to causes of resistance faced by library and information professionals except for R7. It indicates that the professionals have a common consensus over majority of the issues in this part of the questionnaire.

1. There is not much significant difference between the mean scores of the responses elicited by professionals belonging to academic and research groups regarding the statements related to causes of resistance from R1 through R6. There is general agreement upon these.

2. There is a significant difference in responses related to the statement R7. It may be due to nature of their parent organization. Research institutions and organizations are generally well equipped and facilitated. This implies to their support facilities also. Hence their library and information centres are well equipped with required resources when compared to academic set up.

3. There is no significant variance between the mean scores of academic and research library professionals for resistance related statements from R8 to R10.

C2. Comparison of mean scores – Junior vs. Senior

Age Groups: The mean scores of the responses elicited to the statements relating to challenges in the part 1 of questionnaire are tabulated below in two

109




categories viz. Junior Age Group and Senior Age Group followed by its graphical representation for a clear picture.

Table 6: Mean scores of Junior and Senior Age Groups

R1 R2 R3 R4 R5 R6 R7 R8 R9 R10

Junior

Group

(20-40)

1.76

2.03

1.85

2.12

2.21

2.21

1.91

2.12

1.76

1.85

Senior

Group

(40-60)

1.87

1.84

2.07

2.27

2.40

2.22

1.80

2.27

1.85

2.16

Fig. 5 Comparison of mean scores of responses elicited by

LIPs – Age wise Table 6: Z test - Resistance related statements - Age wise

Zc = +1.96 α = 0.05

Junior Group MS

Senior Group

MS J-SD (σJ)

S-SD (σS) Zs Value

Sig / NSig

R1 1.758 1.870

4 0.613

9 0.891

2 -0.76 NS

R2 2.03 1.836

4 0.918 0.876

9 1.05 NS

R3 1.848 2.074

1 0.712

4 1.043

4 -1.27 NS

R4 2.121 2.272

7 0.892

9 1.113

1 -0.71 NS

R5 2.212 2.4 1.023

4 1.098

8 -0.84 NS

R6 2.212 2.218

2 0.992

4 1.030

8 0.00 NS

R7 1.909 1.8 0.842

7 0.779

4 0.58 NS

R8 2.121 2.272

7 0.992

4 1.079

3 -0.68 NS

R9 1.758 1.854

5 0.791

8 0.911

2 -0.55 NS R10 1.848

2.1636

0.8337

0.9956 -1.68 NS

R1 to R10 – Attributes related to causes of resistance in part 2 of questionnaire Junior Group MS (x1) – Mean Scores of Junior Group Responses Senior Group MS (x2) – Mean Scores of Senior Group Responses J- SD (σJ) – Standard Deviation of Mean Scores of Academic Sector Responses S-SD (σS) - Standard Deviation of Mean Scores of Research Sector Responses ZC – Z Critical Value ZS – Z Static Value (Calculated) α − Level of significance of difference in variance

Sig / NSig – Significant / Not Significant From the above table, it is evident that there is no

significant difference between the observed sample means of the two groups based on age. However, inspite of general agreement with respect to the causes of resistance to change, the respondents also suggested some measures for a smooth transition. These suggestions were analyzed in detail using content analysis method. Major suggestions that emerged related to administrative support, employee development programmes, lifelong learning, and personal attributes etc.

Fig. 6 Content Analysis of Suggested Measures

The measures for improvement as suggested by the

respondents to face the challenges and manage resistance indicate requirement of more attention in certain areas. These are management support and administrative measures, promoting employee development programmes, readiness for lifelong learning, improving upon personal attributes of employees, change management related measures like change orientation, readiness for change, resistance management. Also the Information & Communications Technology (ICT) related measures, honing management skills, improving work culture, networking and partnership initiatives, changes in LIS curriculum, service orientation and customer orientation also need a significant attention.

The following four areas discussed here showed a

good precedence over the rest. These are related to administrative measures, employee development, lifelong learning and personal attributes. Amongst the suggestive measures, administrative measures and Employee Development Programmes gained larger consensus over all others. 54 suggestions for administrative measures and 53 suggestions for Employee Development Programme were received.

110




In administrative measures, respondents suggested that there is a need for empowering library and information professionals with decision making positions in organizations. They also suggested that the management should positively provide incentives to LIPs. Such steps by management or administration will motivate the employees to take up challenges and initiate new / customized services to the patrons as well as perform well in their jobs. Suggestions were also received in terms of promoting participative management, skill based job assignment to employees, etc.

To meet challenges, pressure from authorities to perform better and also to meet user expectations a systematic employee development programme is absolutely essential. This will not only help employees to integrate with the vision, mission, and key strategic directions of the library and organization, but also help the library towards becoming a learning organization because “learning organizations are skilled at creating, acquiring and transferring knowledge, and at modifying behaviour to reflect new insights” [5]. Next higher score was for lifelong learning with 39 suggestions. The attribute of lifelong learning can be associated with the concept of learning organization because organizations are made of up of employees. Organizations can become learning organizations if its employees adopt lifelong learning attitude.

Thirty five (35) suggestions were received under the category of personal attributes. They include attitudinal change, readiness to face challenges boldly, inculcate reading habit themselves to promote good reading habits in users, learning from the changing environment, proactive nature, self-motivation, work commitment, and welcoming criticism. These attributes can be either self-developed or gained through training. It is desirable that the professionals improve and sharpen their personal attributes as they prove to be of help during chaotic situations also.

Systems Approach in Library & Information Services: Academic institutions and research organizations can be viewed as complex knowledge systems consisting of several interconnected subsystems that work in synergy and harmony with each other in order to achieve this vision and mission. One of these subsystems is the library, where every LIP can be viewed as a systemic entity in himself.

Organization

The immediate operational environment of LIPs is

impacted upon by changes in the external environment. Thus, the challenges faced by them and their subsequent resistance to them are a result of these changes. Owing to the systems nature of libraries and LIPs, adopting systems thinking is recommended to be adopted in order to face these challenges and successfully dealing with resistance.

Fig. 7 Organizational System

Fig. 8 Operational Environment of LIPs

According to systems thinking approach each entity in

a system is interconnected with every other entity and the behaviour of the system as a whole is shaped by the interconnectedness of these entities. It also advocates that the whole is bigger than the sum of individual entities [6]. Therefore interconnectedness can be achieved by not isolating oneself from the system and by viewing oneself as a part of it i.e., to see the “big picture”. Some of the previous studies conducted using systems approach, are as follows.

• Peter Checkland [7] reports application of Soft Systems Methodology (SSM) in achieving desirable

External Environment

111




and feasible change in the information and library function of a U.K. based company in a science based industry. SSM uses systems models in an organized learning system to improve and make more apparent the process by which ‘what obviously ought to be done’ emerges for particular people at a particular situation with its own unique history.

• John Knowles [8] presents how a soft-systems analysis was carried out for a CD-ROM network for a multi-site polytechnic at Liverpool Polytechnic Library Services. The project involved looking at effects of current situation on the users, in terms of antagonism between different types of users, antagonism between library staff and management, and the general increase in pressure on staff dealing with the running of the current CD-ROM system. The study concludes with a system specification for the recommended CD-ROM network.

• Sulaiman Al-Hassan and A.J. Meadows [9] studied application of soft systems methodology for improving library personnel management in Kuwait. Their study conducted between 1989-91 considering the pre and post Gulf war issues in Kuwait’s libraries, focused on the investigation of experiences of recruitment, training, appraisal, job satisfaction and conditions of service. They have stated that the “study has shown that soft systems methodology can be usefully applied to the examination of library personnel management”.

• Gilbert Tan [10] explains how managers can foster and sustain creativity in their organizations using a total systems approach. He depicts the organization as having four subsystems viz., culture, techno-structural subsystems, management and people. Each system will create unique barriers to creativity. The suggested framework identifies three types of interventions – cultural, organization and design, and training development that can help the ingredients of creativity, foundations, competencies, and support.

• Christian Boissonnas [11] addresses the issue of technical services to re-emphasize it to be a reader service, which is often seen as something other than a reader service. According to this study, the implementation of digital libraries cannot be successful without a comprehensive system-wide approach that calls for people with different experiences and expertise to work together across, rather than in, functional gropus. This systems approach assumes a level of organizational readiness that can be achieved through the deep integration of separate functions.

• Giesecke, J. and McNiel, B. [12] explore in their article why organizations consider attempting to become learning organizations. They include an overview of the theory of learning organizations, present steps to becoming one, and describe examples of learning organization efforts at the University of Nebraska-Lincoln Libraries and other libraries. They adopt Peter Senge’s principles which include ‘systems thinking’.

• Somerville, M.M. and Mirjamdotter, A. [13] present an applied model for cultivating ‘better thinking’ for working smarter within dynamically changing information organizations through systems thinking and application of Soft Systems Methodology (SSM).

• Somerville, M.M. [14] and others report application of Soft Systems Methodology (SSM) tools and information literacy principles and practices to advance data driven dialogue on design and development of enhanced information and knowledge management tools at California Polytechnic State University, California. They report about the library practitioners’ increased confidence and capability to predict productivity enhancement and continuous learning as they assume new roles as architects of digital information and knowledge learning spaces.

• Delbridge, R. and Fisher, S. [15] provide an overview of soft systems methodology (SSM) and review the ways in which the methodology has been applied by managers and researchers to gain a broad understanding of library and information service activity.

Systems Approach for Dealing with Resistance to Change: On similar grounds, systems approach can be applied to managing change in library and information centres in India also. Systems thinking can be applied for dealing with resistance in LIPs, as detailed further. Systems’ thinking allows library and information work to be effective and innovative because library is a part of bigger environment and is interconnected with other departments and units, it cannot function as an isolated entity. LIPs should view themselves as an integral part of the organizational system. In response to the changing environment, there is a need to develop attitudes and competencies in such a manner that they correspond to the changing requirements of the organization, changing technology and user expectations. When LIPs take this view, it helps them to situate library as proactive and not reactive to changes. This requires strong and effective leadership.

112




Effective library leaders should possess qualities like self-awareness, embracing change, customer focus, collaborative spirit, courage and truthfulness, vision, dreams, creativity, innovation and entrepreneurship, planning, trust, values, passion for work, caring for colleagues, communication, transformation and inspirational motivation. Though there can be only one effective leader, yet it is desirable that all LIPs imbibe these leadership qualities so that they have a better understanding of the systemic requirements.

Fig. 10 Causes, Capabilities & Corroboration Model (3Cs

or CCC Model) Figure 10 explains systems approach to dealing with

resistance in LIPs. The impeding technological impacts and ever growing user expectations from LIPs for seamless service are placing heavy pressure on the LIPs to change. By this way both these factors are forming the causes of change. Lack of competencies and negative attitudes among LIPs towards change cause resistance among them. Also the lack of support from authorities in terms of resources and infrastructure contribute further to this resistance.

In such a condition, the LIPs need to develop their capabilities in terms of competencies and attitudes. This can be achieved by encouraging them to undergo continuous professional development and imbibe lifelong learning attributes. Any action requires a two way

approach. Positive stand from LIPs for change will enable the authorities also to offer continuous support in terms of corroboration for their development and better services thereof. Some of the important recommendations are as detailed further. The following competencies and attitudes are recommended to be developed by LIPs.

• Competencies: Library and information professionals require two types of competencies in this changing environment. They are – Professional and Personal. ‘Professional Competencies’ in LIPs are:

o To align with organizational vision, mission and also with key stake holders.

o To assess and communicate the value of information organization. This competency can be linked to knowledge management related measures.

Fig. 9 Systems Approach to Meet Various Challenges o To build a dynamic collection of information resources based on deep understanding of users. Thus, suggestive measures received pertaining to service and customer orientation can be leveraged in terms of developing innovative methods to restructure and customize services as per customers’ requirements.

o To develop and maintain a portfolio of effective and aligned information services.

o To acquire knowledge of new concepts and products or enhanced information solutions. This relates to implementation of ICT and management skills.

‘Personal Competencies’ in LIPs are: o To build and improve upon personal attributes,

which relate to implemeting suggestive measures of developing personal attributes like attitudinal change, being proactive, self-motivation, work commitment, welcoming criticism etc.

o To see the “big picture”, thus enabling the LIP to understand that he is a part of the bigger system and not an isolated entity.

o To create partnerships and alliances. This realtes to implemeting measures related to networking and partnerships like interaction on professional platforms, resource and knowledge sharing, professional networking, and with other departments.

o To employ a team approach; recognize the balance of collaborating, leading and following.

In order to understand continuously happening changes and ever increasing user expectations, LIPs should make a habit of taking feedback from their customers with respect to their service offerings. This will help them stay relevant even in turbulent times.

• Attitudes: The attitudes that are required to be developed by LIPs pertain to that of learning and service. o Learning: Learning can be of two types –

workplace learning and lifelong learning. It is recommended that the LIPs should possess a zeal

113




for learning. LIPs should not only value the opportunity to learn for themselves but also encourage the same opportunity for those around them. Learning takes place in a variety of methods. It can be through enhancing professional qualifications, continuing education, scholarly research and communication, participating in training programmes, workshops, seminars, conferences. Learning also takes place through discussions over professional platforms, exchange of views, experiences, ideas and knowledge sharing. Learning can be augmented by systematic personal knowledge management, by adopting and using emerging technologies to the optimum.

o Service Orientation: It is recommended that every Library and Information Professional should strive towards rendering quality service in spite of the challenges brought about by changes in the information environment. The concept of duty and service inspired by professional values and a desire to serve users better should be the guiding light for all LIPs.

Even in this digitized environment of virtual libraries and repositories, there is still no better theory of library or the role of librarians than the universally incontrovertible and applicable Five Laws of Library Science enunciated by Dr. S. R. Ranganathan [16], that form the bedrock of library and information profession. These are profound yet simple principles based upon linking people, libraries and information they use. In case of any change or transformation, it always holds true to check and recheck if the change or transformation satisfies these laws, provided the words ‘book’ and ‘reader’ are not taken too literally and read as ‘user’ and ‘information’. They are: 1. Books are for use; 2. Every Book – Its Reader; 3. Every Reader – His Book; 4. Save the time of reader; 5. Library is a growing organism. Adhering to these axioms always will help the LIPs to serve their users in a better manner. Supporting Facilities & Environment: As systems approach advocates that the whole is bigger than the sum of its individual entities it is desirable that the management or higher authorities or administration acknowledge that the organization is a system consisting of several interconnected subsystems and therefore extend required support in terms of participative management, empowering employees for decision making, infrastructural facilities, IT related facilities, incentives, developmental programmes etc.

Working environment plays a significant role in helping the LIPs adopt and adapt to changes. A conducive work environment will act positively on their psychology and motivate them to acquire desired attitudes and competencies, thus giving efficient output. Hence it is a prime responsibility on the part of the administration to provide the LIPs a suitable and conducive working environment. Comprehensive Employee Development Programme: An Employee Development Programme with

comprehensive learning domain in thrust areas viz., Information & Communication Technologies, effective usage of various electronic and digital resources, management skills, marketing, strategic planning, total quality management, leadership skills, corporate storytelling for library managers and leaders, organizational learning, building learning organizations, knowledge management etc. is the need of the hour. Focus on organizational communications, service provision, and culture building encompassing customer service and employee relations, supervision and team building, personnel related subjects like legal and effective performance appraisals, performance management should also be developed and initiated for the library and information professionals at all functional levels. Modules related to stress management, self development and ergonomics and healthy working conditions also can be included depending upon the need. Such programmes provide better understanding of organizational development concepts to the LIPs and will help them in developing both competencies as well as attitudes. 5. CONCLUSION Instead of searching for solutions in the environment and expecting support in terms of extrinsic factors, application of ‘changing self to change others and in turn the system as a whole’ approach is desirable. This will lead to a positive change in the attitude of the professionals, which in turn will lead him to contribute in achieving organizational excellence. As nature mirrors systems in each and every of its parts, so also every individual and organization mirror a system in their own way. Realization of existence of these systems to harness their potential to their optimum is needed. REFERENCES [1] L. Pugh, “Change Management in Information

Services”, 2nd Ed., Ashgate, Aldershot, 2007. [2] C. Cardina and D. Wicks, “The changing roles of

academic reference librarians over a ten-year period”, Reference and User Services Quarterly, vol 44(2), pp. 133-142, 2004.

[3] M. G. Melchionda, “Librarians in the age of Internet: their attitudes and roles – a literature review, New Library World, vol. 108 (3/4), pp. 123-140, 2007.

[4] F. Ferreira and Oth., “Information professionals in Brazil: core competencies and professional development”, Information Research, vol. 12 (2), http://InformationR.net/ir/12-2/paper299.html2007.

[5] D. A. Garvin, “Building Learning Organizations”, Harvard Business Review, Pp. 78-91, July-Aug, 1993.

[6] P. Senge, The Fifth Discipline: The Art and Practice of the Learning Organization, Doubleday/Currency, New York, 1990.

[7] P. Checkland, “Achieving 'Desirable and Feasible' Change: An Application of Soft Systems Methodology”, The Journal of the Operational Research Society, Vol. 36, No. 9, Systems Thinking

114




in Action. Conference at Henly, April 1985 (Sep., 1985), pp. 821-831.

[8] J. Knowles, “A Soft Systems Analysis of a CD-ROM Network for a Multi-site Polytechnic”, Journal of Librarianship and Information Science, vol. 25(1), pp. 15-21, 1993.

[9] S. Al-Hassan and A. J. Meadows, “Improving library personnel management: a case study of Kuwait”, Library Management, vol. 15(1), pp. 19-25, 1994.

[10] G. Tan, “Managing creativity in organizations: a total systems approach”, Creativity and Innovation Management, vol. 7(1), pp. 23-31, 1998.

[11] C. M. Boissonnas, “Technical services: the other reader service”, Portal: Libraries and the Academy, vol 1(1), pp. 33-46, 2001.

[12] J. Giesecke and B. McNeil, “Transitioning to the Learning Organization”, Library Trends, vol. 53(1), pp. 54-67, 2004.

[13] M. M. Somerville and A. Mirjamdotter, “Working smarter: an applied model for better thinking in dynamic information organizations”, Association of College and Research Libraries (ACRL) Twelfth National Conference. Minneapolis, Minnesota, April 7-10, 2005.

[14] M. M. Somerville and Oth., “Systems thinking and information literacy: elements of a knowledge enabling workplace environment”, Proceedings of the 39th Hawaii International Conference on Systems Sciences, January 4-6th2006, Kauai, 2006.

[15] R. Delbridge and S. Fisher, “The use of soft systems methodology (SSM) in the management of library and information services: a review”, Library Management, vol. 28(6/7), pp. 306-322, 2007.

[16] S. R. Ranganathan, Prolegomena to Library Classification, Vol. 1, 3rd Ed., UBS Publishers, New Delhi, 1989.

115




APPENDIX Questionnaire for the Research Topic: Causes of Resistance to Change among Library and Information Professionals in India I. Causes of Resistance (Kindly rate your responses) as Strongly agree (1) Agree (2) Cannot say (3) Disagree (4) Strongly disagree (5)

1. Lack of proper attitude towards change causes resistance.

2. Poorly motivated personnel pose resistance to change. 3. Inadequate training causes resistance. 4. Difficulty in understanding fast and complex changes

in the environment sets off resistance. 5. Deep rooted dogmas and technophobia produce

resistance towards technology coupled change. 6. Underdeveloped competencies cause resistance for

the transformed environment. 7. Unavailability of adequate resources and

infrastructural facilities triggers resistance. 8. Lack of customer orientation creates resistance. 9. Absence of management support causes resistance.

10. Resistance is raised due to absence of supportive / conducive work atmosphere.

II. Suggestions for improvement and smooth transition

a. b. c. d. e.

Name of the respondent (optional): Age:

Designation: Email: Organization:

Professional Qualifications:

116




The Rule Based Intrusion Detection and Prevention Model

for Biometric System

Maithili Arjunwadkar 1, R.V. Kulkarni 21 MCA Dept, Modern College Of Engineering, Pune, India

2 Director, SIBER , Kolhapur, India E-Mail: [email protected], [email protected]

ABSTRACT Modern biometric systems claim to provide alternative solution to traditional authentication processes. Even though there are various advantages of biometric process, it is vulnerable to attacks which can decline it’s security. The intrusion detection is an essential supplement of traditional security system. This security system needs the robust automated auditing, intelligent reporting mechanism and robust prevention techniques. We suggest rule based intelligent intrusion detection and prevention model for biometric system. This model contains a scheduler to prepare a schedule to check different logs for possible intrusions, detectors to detect normal or abnormal activity. If activity is normal then alarming and reporting has been executed. If abnormal activity is found the rule engine fires the rule to detect intrusion point and type of intrusion. The model also contains an expert system to detect source of intrusion and suggest best possible prevention technique and suitable controls for different intrusions. This model is also used for security audit as well as alarming and reporting mechanisms. The malicious activity database is stored for future intrusion detection. To detect source tracking backward chaining approach is used. The rules are defined and stored in the Rule engine of the system. Keywords: Security, biometric process, attacks, intrusion detection, prevention, expert system. 1. INTRODUCTION Modern biometric technologies like biometric based authentication system that uses physiological (e.g. thumb print, retina scan, iris) or behavioral (e.g voice, keystroke, touch) claim to provide an alternative for traditional authentication systems that are based on password (token-based) and key (knowledge based). Biometric process or biometric encryption process is used in two separate modes namely enrollment and authentication process. During the enrollment process, the user’s physiological and behavioral characteristics are captured by the sensor. The different feature extractor or key binding algorithms are used to create biometric template. The template is stored during enrollment process to be compared in the future to the one produced during an authentication process. The stored template & the one produced during authentication process is compared by matching algorithm that produces matching result (response Yes/NO). The matched response is then sent to the application, on which a decision algorithm is implemented for granting or denying to the user. This paper is divided into three primary areas. The first section provides an overview of Biometric process, possible intrusions in the biometric process and intrusion detection fundamentals. The second section describes the architecture of intrusion detection & prevention system for intrusions in biometric process. The third section provides how intelligent models are used in this architecture. Finally, the future research and conclusion of intrusion detection & prevention system are presented.

2. RELATED WORK Even though there are various advantages of biometric process, it is vulnerable to attacks, which can decline it’s security. Ratha et al. [15] analyzed these attacks and grouped them into different attack points which are shown in figure 1 [16].

Figure 1: Location of biometric process Attack points According to common criteria of biometric evaluation methodology supplement [16], it is particularly important to consider that attacks can be done on the direct input and output of a biometric template. The Biometric

117






templates are considered to be very sensitive information. They identify and are bound to people. It is the template that is used to determine the user’s rights and privileges to access a resource. When performing a vulnerability evaluation of a biometric system, we must consider a wide variety of generic possible attacks or threats to the security of the system. All elements of a biometric system are susceptible to these threats to some degree. Fig 1 showing locations within the biometric system identified numerically. Throughout the process, there are different points of possible attacks. Some of the possible attack points may be at biometric device, extractor, on channel on which template is transported or transmitted, Extraction/ Comparison unit, Extraction/ Template Storage points during Enrolment, Template Storage units, Template Retrieval unit etc. Before created biometric template, the biometric sample which is really bound to the credentials, privileges, rights, etc. are in most vulnerable state. An attacker may try to substitute his/her own sample or biometric template to masquerade as the intended user. When a template is not associated from its binding with the user, there is the possibility of a substitution attack. If the unbound template is transported or transmitted through an accessible, unprotected medium, then an appropriate means of protection must be considered. The possibility of somehow duplicating the device specific format of the biometric must also is considered for evaluation. This must be done through the analysis of the device algorithms that transform the biometric sample into the template used by the device for comparison, determining the output of the algorithm and then determining the likelihood of duplicating the output through some logic. The different locations which are shown in figure 1 require some detection technique which can be used to detect the attack at those points. The intrusion detection is a necessary supplement of traditionally security protection measures such as firewalls, data encryption, because it can provide real protection against internal attacks, external attacks and abuse [4]. We can incrementally improve security through the use of tools such as Intrusion Detection System (IDS). The IDS approach to security is based on the assumption that a system will not be secure, but that violations of security policy (intrusions) can be detected by monitoring and analyzing system behavior. [14] IShahbaz Parvez et al. [7] describe that, the Intrusion Detection system can be categorized into Anomaly Detection, Misuse detection & hybrid detection which is nothing but combined anomaly & misuse. Anomaly detection is the general category of Intrusion Detection, which works by identifying activities which vary from established patterns for users, or groups of users. Anomaly detection typically involves the creation of knowledge bases which contain the profiles of the monitored activities. Misuse Detection is a second approach to Intrusion Detection. This technique involves the

comparison of a user's activities with the known behaviors of attackers attempting to penetrate a system. Misuse Detection also utilizes a knowledge base of information. Hybrid or Combined Anomaly/Misuse Detection is a third approach, which combine the Anomaly Detection approach and the Misuse Detection approach. The combined approach permits a single Intrusion Detection System to monitor for indications of external and internal attacks. Intrusion detection system which includes functions like monitoring and analyzing both user and system activities and detect suspicious pattern. [7] Most of the intrusion detection systems are available for network security purpose. The biometric process also requires some system which can be used to detect the possible attacks and used some prevention mechanisms to avoid these attacks in future. The authors suggest the Intrusion detection & Prevention System (IDPS) to detect attacks, back tracing the origin of the attack and some suggest prevention mechanisms. 3. ARCHITECTURE OF IDPS To design robust security system, it fulfills the objectives of security like authenticity, confidentiality, integrity, availability & non-repudiation. IDPS (Intrusion detection & Prevention System) contains modules to detect intrusion, filtering intrusion, trace back of intrusion origin, and prevention mechanism for theses intrusions. This security system needs the robust automated auditing & and intelligent reporting mechanism and robust prevention techniques. We suggest security system using intelligent models for biometric protection approach. This system is divided into 3 sub systems :

• Intrusion detection • Backtracking of intrusion source • Prevention techniques.

The components of the intrusion detection and prevention system are shown in figure 2. The Rule based intelligent intrusion detection and prevention model for biometric system contains a scheduler to prepare schedule to check different logs for possible intrusions, and detectors to detect normal or abnormal activity. If activity is normal then standard alarming and reporting would be executed. If abnormal activity is found then the rule engine checks the rule to detect intrusion point and type of intrusion. The model also contains an expert system to detect source of intrusion and suggests best possible prevention technique and suitable controls for different intrusions. This model also uses security audit as well as alarming and reporting mechanisms. The malicious activity database is stored for future intrusion detection. To detect the source by tracking, backward chaining approach is used. The rules are defined and are stored in the Rule engine of the system. Intrusion points & type is passed to expert system. Expert system evaluates that data with known malicious activity database and detects the source using backward chaining approach.

118




Figure 2: Components of Intrusion Detection Process Figure 2: components of intrusion detection and prevention system After detecting source, system suggests the different prevention techniques. For this robust security system the authors use intelligent models like expert system. 4. IDP MODEL AS RULED-BASED EXPERT SYSTEM Expert systems are the most common form of AI applied today in intrusion detection system. An expert system consists of a set of rules that encode the knowledge of a human "expert”. These rules are used by the system to make conclusions about the security-related data from the intrusion detection system. Expert system permits the incorporation of an extensive amount of human experience into a computer application and then utilizes that knowledge to identify activities that match the defined characteristics of misuse and attack. Expert system detects intrusions by encoding intrusion scenarios as a set of rules. These rules replicate the partially ordered sequence of actions that include the intrusion scenario. Some rules may be applicable to more than one intrusion scenario. Rule-based programming is one of the most commonly used techniques for developing expert systems. Rule based analysis relies on sets of predefined rules that can be repeatedly applied to a collection of facts and that are provided by an administrator, automatically created by the system or both. Facts represent conditions that describe a certain situation in the audit records or directly from system activity monitoring & rules represent heuristics that define a set of actions to be executed in a given situation & describe known intrusion scenario(s) or generic techniques. The rule then fires. It may cause an alert to be raised for a system administrator. Alternatively,

some automated response, such as terminating that user’s session, block user’s account will be taken. Normally, a rule firing will result in additional assertions being added to the fact base. They, in turn, may lead to additional rule-fact bindings. This process continues until there are no more rules to be fired. [3]. Consider the intrusion scenario in which two or more unsuccessful authentication attempts are made in a period of time shorter than it would take a human to present biometric info in the login information at biometric sensor. If the rule or rules for this scenario fire, then suspicion level of specific user can get increased. The system may raise an alarm or report ‘freeze action’ to the named user’s account. Account freeze would be entered into the fact database.

Expert system module categorizes the audit data by fact base component initially and then uses relevant detection technique for different audit data. The Rules are defined using JESS. Jess is a clone of the popular expert system shell CLIPS which rewritten entirely in Java. If biometric template stored in central database, alteration and deletion of biometric template is not allowed to any user except root or system or super user for database administration purpose. The attacker modifies or deletes the biometric template. The rule which is to be checked for unauthorized modification of biometric template is: Figure 3: unauthorized Modification of biometric template rule The rule which is to be checked for unauthorized deletion of biometric template is: Figure 4: unauthorized Deletion of biometric template rule The imposter steals the biometric template of an authorized user from template storage or from other biometric system. The rule which is to be checked for illegal copy of biometric template is: Figure5: unauthorized Copying of biometric template rule 5. CONCLUSION In spite of the various advantages of biometric process, it is vulnerable to attacks which can compromise on it’s security objectives. Intruder can attack on different points of biometric process for example biometric

Detector Activit Normal

Activity Data

Normal? Yes

Rule Engine

Alarming & R ti

No

Known Malicious activity

Detected Intrusion

i t &

IF( (user is “root” || “superuser”|| “system”) &&( transaction_type is “Modification”) && ( not(time_stamp is normaltime_stamp))) THEN (Alart: “UnauthorizedModification”)

IF( (user is “root” || “superuser”|| “system”) &&( transaction_type is “Deletion”) && ( not(time_stamp is normaltime_stamp))) THEN (Alart: “Unauthorized Deletion”)

IF (transaction_type is “Copy”)

THEN (Alart “Unauthorized Copying )

119




120

template database, network channel, biometric device, template creation module etc. The model suggested in this paper is useful to detect the intrusion in both types. The model also contains an expert system to detect source of intrusion and suggests best possible prevention technique and suitable controls for different intrusions. This model also uses security audit as well as alarming and reporting mechanisms. The malicious activity database is stored for future intrusion detection. To detect the source by tracking, backward chaining approach is used. The rules are defined and are stored in the Rule engine of the system. The intelligent model uses AI and expert system is backbone of this system. 6. FUTURE Work In this paper the authors design the architecture of the model for intrusion detection and prevention in biometric process. In future the research will expand the different methods for intrusion detection of different vulnerabilities. The backward chaining approach of expert system can be used to detect source of intrusion Also we design some rules for backtracking of source using backward chaining process and suggest different prevention techniques. REFERENCES [1] Computer System Intrusion Detection: A Survey By

Anita K. Jones and Robert S. Sielken [2] Feature deduction and ensemble design of Intrusion

detection system By Srilatha Chebrolu, Ajith Abraham,Johnson Thomas Computer & security (2004)

[3] An Intrusion detection system expert system with fact based ByYuan Yuan,Dai Guanzhong Asian Journal of Information Technology 6(5) 614-617, 2007

[4] The Application of Generic Neural Network in Network Intrusion Detection By Hua Jian, Junhu

Ruan in Journal of computers vol 4 no 12 December 2009.

[5] Design & Implementation of rule based expert system for fault management, by Su Myat Soe and Paing Paing Zaw in World academy of science , engineering & technology 48 2008

[6] Jess in Action by Ernest Friedman -Hill [7] A Comparative Analysis of Artificial Neural Network

Technologies in Intrusion Detection Systems by Shahbaz Pervez, Iftikhar Ahmad, Adeel Akram, Sami Ullah Swati

[8] Intelligent system for information security Management: architecture & design issues by Marina Hentea in Informing science & information technology vol. 4 2007.

[9] Neural Network Learning based chaos by Truong Quang Dang Khoa & Masahiro Nakagava in International Journal of computer & system Science & engineering 1:2 2007

[10] One way hash functions based on neural network Shiguo Lian,Jinsheng sun, Zhiquan Wang

[11] Password based a generalize robust security system design using neural network By Manoj Kumar Singh In IJCSI international Journal of computer science issue vol 4 no 2,2009

[12] An Introduction to Intrusion Detection by Aurobindo Sundaram

[13] On artificial Intelligence approaches for network Intrusion detection systems by sattar B. Sadkhan in MASUAM Journal of computing vol. 1 issue 2 , September 2009.

[14] A Backpropagation Neural Network for Computer Network Security Khalil Shihab

[15] “An analysis of minutiae matching strength” by N.K.Ratha , J.H. Connell and R.M. Bolle Proc AVBPA 2001, Third international conference on Audio and video based biometric person authentication pp 223-228 2001.

[16] Biometric Evaluation Methodology (BEM) suppliment, Produced by the common criteria Biometric evaluation methodology working group. Version 1.0 August 2002.

VOL. 1, NO. 2, Oct 2010 ISSN 2079-8407 Journal of Emerging Trends in Computing and Information Sciences



121

Using Multimodal Fusion in Accessing Web Services

Atef Zaguia1, Manolo Dulva Hina1,2, Chakib Tadj1, Amar Ramdane-Cherif2,3

1LATIS Laboratory, Université du Québec, École de technologie supérieure 1100, rue Notre-Dame Ouest, Montréal, Québec, H3C 1K3 Canada

2PRISM Laboratory, Université de Versailles-Saint-Quentin-en-Yvelines 45, avenue des États-Unis, 78035 Versailles Cedex, France

3LISV Laboratory, Université de Versailles-Saint-Quentin-en-Yvelines 45, avenue des États-Unis, 78035 Versailles Cedex, France

Email : [email protected]

ABSTRACT In our days, the technology allows us to produce extended and totally human-controlled multimodal systems. These systems are equipped with multimodal interfaces allowing a more natural and more efficient interaction between man and machine. End users could take advantage of natural modalities (e.g. eye gaze, speech, gesture, etc.) to communicate or exchange information with applications. The use of multimodal applications in web services, integrated with natural modalities, is an effective solution for users who cannot use a keyboard or a mouse, on users who have visual handicap, on mobile users equipped with wireless telephone/mobile devices, on weakened users, etc. Our work presents an approach in which various input modalities (speech, touch screen, keyboard, eye gaze, etc.) are available at user’s disposition in order to access web services. While current state-of-the-art uses two (on rare cases, three) pre-defined modalities, our approach allows an unlimited number of concurrent modalities using semantic level called “multimodal fusion”. Such approach gives user the flexibility to use the modalities as he sees fit for his situation. The detailed description of the proposed approach as well as the application that has been developed that uses these multimodalities is presented in this paper. Keywords: multimodal fusion, web service, human-computer interface. 1. INTRODUCTION

Ever since the computer was born, one of the biggest challenges in informatics has always been the creation of systems that allow transparent and flexible human-machine interaction (Sears and Jacko 2007; Aim, Alfredson et al. 2009). Since the 1960’s, computer evolution has been rapid and as always, the goal has been to satisfy the needs of users and come up with systems that are intelligent, more natural and easier to use.

Various researches have been directed towards the creation of systems that facilitate communication between man and machine (Yuen, Tang et al. 2002) and allow a user to use his natural modalities (eye gaze, speech, gesture, etc.) in communicating or exchanging information with applications. These systems receive inputs from sensors or gadgets (e.g. camera, microphone, etc.) and make an interpretation and comprehension out of these inputs; this is multimodality (Ringland and Scahill 2003; Ventola, Charles et al. 2004; Carnielli and Pizzi 2008; Kress 2010). A well-known sample of these systems is that of Bolt’s “Put that there” (Bolt 1980) where he used gesture and speech to move objects.

In our days, various multimodal applications (Oviatt and Cohen 2000; Yuen, Tang et al. 2002) have been conceived and are effective solutions for users who, for one reason or another, cannot use a keyboard or a mouse (Shin, Ahn et al. 2006), on users who have visual handicap (Raisamo, Hippula et al. 2006) , on mobile users equipped with wireless telephone/mobile devices (Lai,

Mitchell et al. 2007), on weakened users (Debevc, Kosec et al. 2009), etc. These applications are integrated with web services which are software components that represent an application function or service. The web services can be accessible from another application (a client, a server or another web services) within the Internet network using the available transport protocols (Li, Liu et al. 2007). This application service can be implemented as an autonomous application or a set of applications. It pertains to a technology allowing applications to communicate with one another via Internet, independent of the platforms (i.e. Windows (Microsoft 2010), Linux (Red_Hat_Enterprise 2010), Mac OS (Apple 2010), etc.) and languages (i.e. Java (Bates and Sierra 2003), C (King 2008), C++ (Malik 2010), J2EE (Weaver and Mukhar 2004), etc.) that these applications are based.

Our approach is to develop a flexible system based on application, capable of manipulating more than two modalities. The approach consists of modules that detect the modalities involved, take into account each modality’s parameters and perform the fusion of the modalities in order to obtain the corresponding action to be undertaken. This approach is more flexible than current state-of-the art systems that run on predefined two modalities. The fusion of modalities involved in this work is based on the modalities selected by the user as he sees suitable to his situation.

The rest of this paper is organized as follows. Section II takes note of other research works that are related to





122

ours. Section III discusses the modalities and multimodal fusion system, section IV is about the discussion on the components of the multimodal fusion and their specification. The paper is concluded in section V; it also takes note of other works that we intend to do in the near future.

2. RELATED WORK

Modality refers to the path or channel by which human and machine interact with one another. An impoverished traditional computing set-up uses mouse, keyboard and screen by which the user interacts with the computer and vice-versa. The concept of multimodality allows the use of other means, or modality, in sending user input to the machine as well as the machine sending output to the user. Such modalities include other gadgets and sensors, such as touch screen, stylus, etc. and man’s natural modalities, such as speech, eye gaze and gestures. The invocation of multimodalities permits a more flexible interaction between user and machine as well as allowing users with temporary or permanent handicap to benefit from the advancement in technology in undertaking computing tasks. Multimodality allows the invocation of other means when some other modalities are not available or possible to use. For example, speech (Schroeter, Ostermann et al. 2000) can be a more effective input modality than a mouse or a keyboard if the user is on the go. Using multimodality in accessing user application is an effective way of accomplishing user’s computing task.

Web service (Ballinger 2003) is a software component that represents an application function (or application service). It is a technology that allows applications to interact remotely via Internet, independent of platforms and languages on which they are based. The service can be accessed from another application (a client, server or another Web service) through Internet using transport protocols. Web services are based on a set of standardizing protocols. Theses protocols are divided in to four areas, namely: the transport layer, the XML messages, the description of services and the search service.

Accessing web services using various modalities has been done and implemented in many applications for various services. For example, the work of (Caschera, D'Andrea et al. 2009) presents an effective web-based multimodal system (Madani, Nigay et al. 2005) that can be used in case of disasters, such as earthquake. In that work, the user sends and receives information using his mobile device through voice, touch or eye gaze (Zhang, Imamiya et al. 2004). This multiple-modalities system ensures simplicity of use and instant access to information that are pertinent in providing emergency services in critical situations, and consequently, save many lives. The work of (Steele, Khankan et al. 2005) demonstrates the concepts of discovery and invocation of services. The concept is illustrated using an example, that of a passenger in an airport. Here, the passenger can use his cell phone to know the available services in the airport. Using voice and

touch, the user can browse and select his desired services. This system is significant as it satisfies the user’s urgent needs for easy and quick access to information. In (Pfleger 2004), the author presents a system commonly used in house construction, namely the bathroom design. The multimodal system interface (Oviatt 2002) spontaneously integrates speech and stylus inputs. The output comes in the form of voice, graphic or facial expressions of a talking head displayed on screen. Here the user interacts with the system to get an ongoing assistance during the bathroom design. (Giuliani and Knoll 2008) presents a case of a human-robot multimodal interaction. Here, the two-armed robot receives vocal and non-verbal orders to make or remove objects. The use of such robots with remote control can be very beneficial especially in places where access is dangerous for human beings. In (Wang, Zhang et al. 2006), the authors proposed a multimodal system that helps children learn the Chinese language through stylus and voice. The system presents a good way for educating children as it helps them learn while playing.

In a nut shell, it could be said that the above- mentioned multimodal systems are of great importance to the users in terms of efficiency. However, the very fact of their being based on only two (and on rare occasion, three) modalities makes the use of most of these systems challenging for users. This leads us to the creation of a multimodal system with an unlimited number of modalities in order to provide an easier interface access and simpler usage for the users.

3. MODALITIES AND MULTIMODAL FUSION SYSTEM

In this section, we describe the approach that we propose and all other modules involved in the implementation of such approach.

A. Architecture and Approach

Accessing a web service involves the use of four web-service modules. These modules need to be loaded on a computer, on a robot, or any machine that can communicate via Internet or social network. The architecture of our proposed system is shown in Figure 1.




123

Figure 1. Architecture of multimodal fusion system for accessing web services

As shown in the diagram, the system consists of the following elements: • Parser – it takes an XML file as input, extracts

information from it and yields an output indicating the concerned modality and its associated parameters

• Parameter Extractor – the output from the parser serves as input to this module, it then extracts the parameters of each involved modality

• Fusion and Multimodality – based on the parameters involved in each modality as well as the time in consideration, it decides if fusion is possible or not

• Internet/Social Network – serves as the network by which the user and the machine/computer involved communicate

• Computing Machine/Robot/Telephone – this is the entity by which the user communicates with

The processes involved in the multimodal fusion are shown in Figure 2. Two or more modalities may be invoked by the user in this undertaking. Consider for example the command “Replace this file with that file” wherein the user uses speech and a mouse click to denote “this file” and another mouse click to denote “that file”. In this case, the modalities involved are: input modality 1 = speech and input modality 2 = mouse. The processes involved in the fusion of these modalities are as follows: • Recognition – this component converts the activities

involving modalities into their corresponding XML files.

• Parser Module, Parameter Extraction Module and Multimodal and Fusion Module – same functionalities cited earlier

• Action – this involves the corresponding action to be undertaken after the fusion has been made. The resulting output may be implemented using output modality 1, 2, …, n. In the same case cited earlier, the output modality involved is the screen. It is also possible that the confirmation of such action may be presented using a speaker.

• Feedback – when conflict arises, a user receives a feedback from the system. For example, in the cited case, if “this file” and “that file” refer to the same entity, the user is informed about it via feedback.

Figure 2. Framework of multimodal fusion

All of these modules may have been installed or are situated in any location within the network. The modules themselves communicate with one another in order to exchange information or do a task such as instruct a robot to do something.

B. Multimodal Fusion

Fusion (Pfleger 2004; Pérez, Amores et al. 2005; Lalanne, Nigay et al. 2009) is a logical combination of two or more entities, in this case two or more modalities. Modality signals are intercepted by the fusion agent and then combine them based on some given semantic rules.

Based on literature review, two sets of fusion schemes exist: the early fusion and the late fusion (Wöllmer, Al-Hames et al. 2009). Early fusion (Snoek, Worring et al. 2005) refers to a fusion scheme that integrates unimodal features before learning concept. The fusion takes effect on signal level or within the actual time that an action is detected (Oviatt, Cohen et al. 2000). On the other hand, late fusion (Mohan, Dhananjaya et al. 2008) is a scheme that first reduces unimodal features to separately learned concept scores, and then these scores are integrated to the learned concepts. The fusion is effected on semantic level. In this work, the fusion process used is the late fusion. Assume for instance the arrival of modality A, along with its parameters (e.g. time, etc.) and another modality B with its own parameters (e.g. time, etc.), then the fusion




124

agent will produce a logical combination of A and B, yielding a result, we call C. The command/event C is then sent to the application or to the user for implementation. The multimodal fusion can be represented by the relationship f: C = A + B.

Figure 3 describes the steps involved in the fusion process. As shown, an n number of XML files involving modalities serve as input to the system. Recall that each XML file contains data related to modalities as well as their corresponding parameters. The steps undertaken are as follows:

Figure 3. The detailed fusion process

• Step 1: The system begins asking if the scenario (i.e. the condition) is already in the database. If so, we already know the result (also stored in the database); it is retrieved and is to be implemented. Otherwise, step 2 is performed.

• Step 2: This indicates that the scenario is new to the system. The system asks if the semantic of the operations just performed or about to be performed using the involved modalities are equivalent. Consider for example a case in which, using speech, the user says: “Write 5” and also using keyboard, he types “5”. Technically, the two actions mean the same and are therefore semantically equivalent. The result means that there is no fusion to be made. The information will be stored in the database for reference in case such event happens again in the future. The desired action is that only one of these commands – one or the other – will be implemented.

• Step 3: There is a conflict if two or more commands are in contradiction with one another. For example, if the user, using speech, says: “Write 5” and using stylus, for example, he writes “4”.

• Step 4: A feedback is sent to the user to correct a detected conflict. If the user takes no action, the system

continues on sending user a feedback for the resolution of the problem. The user needs to decide which of the two commands (or n commands, in general) – one or the other – must be considered by the system for implementation.

• Step 5: If the scenario is completely new and there is no conflicting data and action involved, all the corresponding data will be stored in the database.

• Step 6: Queries are sent to the database by the fusion agent to retrieve information involving modalities. The fusion is then performed and the result is also stored in the database.

• Step 7: Result refers to the desired action to be performed using the modalities involved. As shown in Figure 4, the fusion agent is composed of

three sub-components, namely: • Selector – it interacts with the database in selecting the

desired modalities. It retrieves 1 .. m modalities at any given time.

• Grammar – verifies the grammatical conditions and all the possible interchanges between the modalities involved.

• Fusion – this is the module that implements the fusion function.

Figure 4. The fusion agent

The algorithm of the fusion process is shown in Figure 5. The essence of time is important in determining whether the actions involved concerning modalities are qualified for fusion not.

The process begins with accessing the database and taking the data associated with one modality (step 1). The next step is to check the grammatical implication of the associated modality (step 2). Here, the implication analysis could determine if the action to be taken involving such modality could be interpreted as a unimodal or multimodal action (needing complementary data from another modality). For example, the speech command “Put that there” implies that some data associated with “that” and “there” are expected. On the other hand, a speech command such as “Draw a box” implies that the command can stand on its own – its meaning is complete, needing no complementary data – and therefore should be taken as a




125

unimodal activity. If the desired action requires multimodal inputs, then fusion is performed whereas if it is a unimodal one, no fusion should be performed. In both of these cases, their results are stored in the database for future reference. As the diagram shows, the fusion process is a continuous process, always in effect while the user is logged on into the computing system.

Figure 5. The fusion algorithm

Failure in grammatical conditions may also arise. For

example, a vocal command “Put there” is a failure if there is no other complementary modality – such as touch, eye gaze, mouse click, etc. – is associated with it. If such case arises, the system looks at some other modalities that come within the same time interval as the previous one that was considered. C. Time Element in the Fusion Process

Here, we shall illustrate various variations of time frame which will decide if fusion is possible or not. For all the cases involved, the following notations shall be used: Mi = modality i, t1Mi = arrival time of modality i, t2Mi = end time of modality i, tmin = minimum time delay allowed and is dependent of the modality involved. With reference to Figure 6, here are some of the possible cases:

Figure 6. The time element involved in two modalities

• Case A: Modality 1 arrives at t1M1 and ends at t2M1. Another modality 2 arrives at t1M2 and ends at t2M2. They, however, occur on different time frame and are mutually exclusive of one another. As they are independent of one another, each modality is unimodal. Each modality is treated separately. Hence, f1: C1 = M1 and f2: C2 = M2

• Case B: Modality 1 and modality 2 arrive separately but the arrival time of these two modalities is lesser than tmin. In this case, a test is needed to determine if fusion is possible. In the affirmative case, the resulting function would be: f: C = M1 + M2

• Case C: Modalities 1 and 2 arrive within the same time frame. Here, fusion is obvious and the resulting function is f: C = M1 + M2.

4. COMPONENTS OF MULTIMODAL FUSION AND THEIR SPECIFICATIONS

In this section, we present the different components that are involved in the multimodal fusion process and provide each component’s specification. The formal specification tool Petri Net as well as an actual program in Java are used to elaborate on the system components’ specifications.

A. The User Interface

As true with every computing system, our system has a user interface (Oviatt and Cohen 2000) with which the user can communicate to the computing system. In the user interface, the user can select modalities that he wishes. An event concerning the modality is always detected (e.g. was there a mouse click? was there a vocal input?, etc.). The system keeps looping until it senses an




126

event involving modality. The system connects to the database and verifies if the event is valid. An invalid event, for example, is a user’s selection of two events using two modalities at the same time when the system expects that only one event can be executed at a time. If the event involving modality is valid, an XML file is created for that modality and its associated parameters. The XML file is forwarded to the parsing module. The parser then extracts data from the XML tags and sends the result it obtained to the Multimodal and Fusion module. The diagram in Figure 7 demonstrates this process.

B. The Parser

This module receives as input XML files containing data on modalities, usually captured using sensors (e.g. webcam, micro, touch screen, etc.). From each XML file, this module extracts some tag data that it needs for fusion. Afterwards, it creates a resulting XML containing the selected modalities and each one’s corresponding parameters.

In conformity with W3C standard on XML tags for multimodal applications, we use EMMA. EMMA (Desmet, Balthazor et al. 2005; Johnston 2009; W3C 2010) is a generic tagging language for multimodal annotation. It is an integral part of the W3C norm for multimodal interactions. Its general objective is to automatically represent the information extracted from user inputs through interpretation of system components. The EMMA tags represent the semantically recovered input data (e.g. gests, speech, etc.) that are meant to be integrated to a multimodal application. EMMA was developed to allow annotation of data generated by heterogeneous input media. When applied on target data, EMMA result yields a collection of multimedia, multimodal and multi-platform information as well as all other information from other heterogeneous systems.

Figure 7. The system’s user interface

For instance, assume that the module receives the following XML files in EMMA notation (see Figure 8). The corresponding EMMA tags and their meanings are as follows: • <emma:emma> is the root node, holding EMMA

version and namespace information, and providing a container for one or more of the following elements:

• <emma:interpretation> is used to define a given interpretation of input, and holds application specific markup;

o id is a required xsd:ID value that uniquely identifies the interpretation within the EMMA document.

• <emma:group> is a general container for one or more interpretations. It can be associated with arbitrary grouping criteria; the emma:group element is used to indicate that the contained interpretations are related in some manner.

• The emma:model annotation may refer to any element or attribute in the application instance data, as well as any EMMA container element (emma:one-of, emma:group, or emma:sequence).

• The emma:lang annotation is used to indicate the human language for the input that it annotates. The values of the emma:lang attribute are language identifiers. For example, emma:lang="fr" denotes French while emma:lang="en-US" denotes US English

• The confidence score in EMMA is used to indicate the quality of the input, and it is the value assigned to emma:confidence in the EMMA namespace

• emma:start and emma:end are attributes indicating the absolute starting and ending times of an input in terms of the number of milliseconds

Using speech and touch screen specimen modalities

shown in Figure 8, the resulting combined XML file is similar to the one shown in Figure 9 (Left). Then the fusion of these two modalities yields the result that is shown in Figure 9 (Right). The fusion result indicates that the object cube is moved to location (a,b).




127

Figure 8. Specimen XML files containing modalities

(speech and touch screen)

Figure 9. (Left) The resulting combined XML file and (Right) the resulting XML file after the fusion.

C. The Parameter Extractor

The manipulation of an XML file is a task usually performed within the development phase of an application. It is usually undertaken by a parser. A XML parser is a library of functions that can manipulate on an XML document. In selecting a parser, we usually look for two characteristics – that of parser being efficient and rapid.

The parser used in this system is called DOM (Document Object Model) (Wang, Li et al. 2007). It is a large, complex and stand-alone system that uses object model to support all types of XML documents. When parsing a document, it creates objects containing trees with different tags. These objects contain methods that allow a user to trace the tree or modify its contents. See Figure 10.

Figure 10. DOM parsing XML document

DOM works in two steps. The first step involves the loading of an XML document (this is also when DOM consumes memory space). The second step involves performing different operations on the document. The advantages of using DOM are: (1) the easy traversal of its tree, (2) easy way of modifying the contents of the tree, and (3) traversal of file in whatever direction the user desires. On the other hand, its disadvantages include: (1) large consumption of memory and (2) processing of the document before using it.

Using the same example cited in the previous sub-section, the resulting DOM tree after the parsing process is shown in Figure 11.

Figure 11. A sample DOM tree




128

D. The Database The database stores all modalities identified by the

system and the modalities’ associated parameters. There are many available databases, such as PostgreSQL, MySQL, SQL Server, Oracle, Access, etc. In this work, the database used is PostgresSQL (PostgreSQL 2010). Using PostgresSQL, the parameters, values and entities of the database are defined dynamically as the module parses the XML file.

As shown in Figure 12, our database consists of seven tables, namely: • Modality – this table contains the names of modalities,

the time an action involving the modality begins and the time that it ended.

• Modality_Added_Parameters – this table contains all the attributes of every modality.

• Modality_Main_Parameters – contains the name of all parameters and their values

• Union_Modality_Main_Parameters – this table links the modality and their parameters

• Fusion – this table contains all the fusions that had been implemented. This table allows us to keep the previous historical data that can be used later for learning.

• Fusion_Main_Parameters – contains the names of parameters and their values that are associated with the multimodal fusion.

• Union_Fusion_Main_Parameters – this table serves as a link to the multimodal fusion that was just made, including its corresponding parameters

Figure 12. Tables that make up the database

To demonstrate what data gets stored into the database, let us cite an example using the following XML file (see Figure 13):

Figure 13. Specimen XML file

Hence, given the cited case (Figure 13), the contents of the Modality table are shown in Table 1.

Table 1. A specimen Modality table

Index Name Start_Time End_Time Index_ Fusion

1 Speech 10 :11 :11 10 :12 :13 1 2 Mouse

click 10 :10 :11 10 :12 :13 1

The contents of this table and their meanings are: • Index – this is a redundant element used by database

management for optimizing requests. In our case, it represents the index for each modality.

• Name – this column tabulates all the names of all modalities.

• Start Time – indicates when a modality is started. • End Time – when a modality activity ended. • Index_Fusion: indicates the index of the fusion table.

This column is filled with the necessary data during the fusion process to denote the modalities that have been fused.

The content of the Modality_Main_Parameters table, for the cited case is shown in Table 2. The contents and meanings of each element in this table are as follows: • Index – shows one index for every parameter of a

modality. • Name - provides the name of each parameter. • Value – identifies the value of each parameter.

Table 2. A specimen Modality_Main_Parameters table

Index Name Value 1 Action move 2 object Yellow 3 aim Put in position 4 Position X a 5 Position Y b

Likewise, the Union_Modality_Main_Parameters table contents, for the cited case, are shown in Table 3. The items in this table are:

http://www.postgresql.org/

http://www.mysql.fr/

http://www.microsoft.com/france/serveur/sql/default.mspx

http://www.oracle.com/

http://www.microsoft.com/france/office/2007/programs/access/overview.mspx




129

• Index – same as earlier definition. • Index_Modality – shows the index to the table

“Modality”. • Index_Main_Parameters – shows the index to the table

“Modality Index Parameters”. Table 3. A specimen Union_Modality_Main_Parameters

Index Index_Modality Index_Main_Parameters 1 1 1 2 1 2 3 1 3 4 2 4 5 2 5

The contents of the Fusion table, of the

Union_Fusion_Main_Param table, and that of the Fusion_Main_Param are all shown in Table 4, Table 5,

and

Table 6, respectively. The meanings of the columns in Table 4 are as follows: • Index – same as before. • Name_Fusion – identifies the name of the fusion.

Table 4. A specimen Fusion table

Index Name_Fusion 1 fusion 1 2 fusion 1

The contents of Table 5 are as follows: • Index – same meaning as before. • Index_Param – shows the index of the parameters of

the table “Fusion_Main_Param”. • Index_Fusion - tells us the index number of the

fusions associated to each parameter.

Table 5. A specimen Union_Fusion_Main_Param table

Index Index_Param Index_Fusion 1 1 1 2 2 1 3 3 1 4 4 1 5 5 1 6 6 1

In

Table 6, the meanings of its columns are: • Index – same as before. • Name – present the names of each parameter involved

in the fusion.

• Value – present the values of each parameter involved in the fusion.

Table 6. A specimen Fusion_Main_Param table

Index Name Value 1 action move 2 object cube 3 old position X i 4 old position Y j 5 new position X a 6 new position Y b

In Table 7, the meanings of its columns are: • Index – same as before. • Index_Modality – the data here serves as index to the

table “Modality”. • Name – lists down the name of the attribute of each

modality. • Value – lists down the value of value of the attribute of

each modality. Table 7 contains the attributes of each modality. In the cited case, however, we did not use attributes therefore the table column is empty.

Table 7. Modality_Added_Parameters

Index Index_Modality Name Value1

E. Sample Case and Simulation using Petri Net Here, we will demonstrate a concrete example and

describe its specification using Petri Net. We will briefly discuss the basic concepts of Petri Nets in order that the next diagrams involving it would be clear and easily comprehensible. Petri Net (ISO/IEC-15909-2 2010; Petri-Nets-Steering-Committee 2010) is an oriented graph. It is a formal, graphical, executable technique for the specification and analysis of a concurrent, discrete-event dynamic system. Petri nets are used in deterministic and in probabilistic variants; they are a good means to model concurrent or collaborating systems. They also allow for different qualitative or quantitative analysis that can be useful in safety validation. The symbols and notations used by Petri Net are shown in Figure 14. Places (represented by circles) are states in a simulated diagram whereas transitions (represented by rectangles) are processes that are undertaken by a certain element. A certain element goes from one state to another through a transition. Usually a certain element begins in an initial state (manifested in Petri Net via an initial token in a place). When an element goes from state “a” to state “b” through a transition, it is




130

manifested in the Petri Net via a movement of token from place “a” to place “b” via transition “x”.

Figure 14. Petri Net symbols and notations

In the specifications that will follow in this paper, only a snapshot of one of many possible outcomes is presented; this is due to space constraints in which we do not have the space to show all possible variations of inputs and corresponding outputs. The application software PIPE2 is used in simulating Petri Net. PIPE2 (Bonet, Llado et al. 2010) is an open source, platform independent tool for creating and analysing Petri nets including Generalised Stochastic Petri nets.

As shown in Figure 15, the sample application is about car rental. As shown in Figure 15(A), the menu is composed of four selections – the rental option, the sign-up option, the manage reservation option and the usage option. For simplicity of the discussion, the usage option, as shown in Figure 15(B), allows the user to select and identify his preferred modalities. In this example, we listed 4 specimen modalities, namely: (1) voice, (2) touch screen, (3) keyboard and (4) eye gaze. When the user signs up for a rent, the rental period needs to be specified, hence, in Figure 15(C), the interface allows the user to specify the month, the day, and the time for both the start and end of the rental period. Finally, in Figure 15(D), we provide an interface which allows the user to input his coordinates as well as the credit card information.

To make use of the above-mentioned application, we will cite a sample case. Wishing too know the country better, four students have decided to take a trip in a weekend. They decided to rent a car for them to see some regions of the country. One night, François opened his computer equipped with touch screen and connected himself to a rent-a-car website. Using speech, touch screen, eye gaze and keyboard, he was able to do a car reservation for the following week. During the reservation process, some XML files are created, with different modalities used. These files are sent

to the server of the car rental enterprise within the Internet network using the http protocol. These files are to be sent first to the “Parser” module to extract all the modalities involved. Then this module will create another XML file that contains all the different modalities as well as their corresponding parameters. Then this file will be sent to the “Parameter Extractor” module which will extract all the parameters of the modalities involved and send them to the “Fusion” module. E.1. Scenario

François runs the rent-a-car application software. The initial interface is displayed. Using voice, he selected “Rent”. Then the second interface is presented; using touch screen, he chose “Start Day”. Using keyboard, he types “25”. Then using eye gaze, he selected “Start Month” and via speech, he said “January”. Then using keyboard, he selected “Start Time” and he entered “1:30 pm” using keyboard. Then “End Day” is selected using speech and he uttered “27”. Using keyboard, he selected “End Month” and types in “1”. At the end, using eye gaze, he chose “End Time” and said “6:00 pm”. At the end of the reservation process, he received a confirmation message through his laptop computer. This scenario is depicted in the diagram of Figure 16.

E.2. Grammar The diagram in Figure 17 shows the grammar used for the interfaces A and B (see Figure 15) of the sample rent-a-car application. The choice, for instance, is defined as a selection of one of the menus (i.e. rent, sign up, manage reservation and usage) in the user interface. The choice of time is in American time format (example: 12:30 pm); choice of month can be numeric (e.g. 1) or alphabetic (e.g. January) and the interface2, among others, is defined as a selection of either an entry of month, day or time or a back/next/cancel selection.

E.3. Simulation 1 The diagram in Figure 18 shows the interactions

involved in interface A in which the user would have to choose one option in a rent-a-car menu. The rest of the diagram demonstrates all activities when and after the user chooses “Rent” via speech. As shown, an XML file is created which is then sent to the network. The Parser module parses the XML data and extracts tags that contain modality information including its associated parameters. The parameter extractor module extracts the necessary parameters and is then forwarded to the Multimodal and Fusion Module. As it is a unique action, in this example, no fusion is implemented. It is a unimodal action. Nonetheless, it is saved onto the database and the interface B and all menus associated with the “Rent” option are to be instantiated.

E.4. Simulation 2




131

The diagram in Figure 19 demonstrates a net showing the system activities when the “Rent” option is selected (see the token in the place called “Rent Selected” in the middle of the net). At the same time that this option is selected, the three modalities are also selected (see the tokens in keyboard modality, speech modality and touch screen modalities). The Petri Net diagram shows us all possible variations that could arise. For example, when the keyboard is selected (upper right corner of the net), the next possible transition may be one of the following: (1) Select month by keyboard, or (2) Select Day by keyboard or (3) Time selected by keyboard. The same can be said of the two other modalities – speech and touch screen. Here, our desired output is a data entry for month, day or time which needs to be implemented using only one modality per parameter. For example, month selected by two or more modalities is invalid. In the diagram, a snapshot of one of the many possible outcomes is shown – here, the month is selected by touch screen, day is chosen using keyboard and time is chosen via speech. We colour the states for easy viewing – yellow is associated with speech,

blue for keyboard modality and pink for touch screen modality; the red circle denotes that it is assumed that the “Rent” option is already selected. E.5. Simulation 3

The diagram in Figure 20 shows the intricate Petri Net demonstrating various states and transactions as the user uses four modalities (e.g. keyboard, speech, touch screen and eye gaze) in selecting and entering specimen data for “Start Month”, “Start Day” and “Start Time”. This diagram partly shows the Petri net involved in implementing the activities as shown in Figure 16. Again, for the purpose of simplicity, we put colours on the places of the net: (1) blue – for the multimodal transactions in which touch screen is used to select “Start Day” and keyboard to enter “25”, (2) green – to show the places and transitions involved when user uses eye gaze to select “Start Month” and speech to input “January”, and (3) pink – in which “Start Time” is selected using eye gaze, “1:30” is uttered vocally, and “pm” is entered via keyboard.

Figure 15. (A) Car rental menu, (B) Available modalities, (C) Data entry for month, day and time, and (D) Data entry for

renter’s information




132

Figure 16. A sample scenario showing multimodal interactions between the user and the machine

Figure 17. Grammar for the period that a user would rent a car




133

Figure 18. System activities as the user chooses “Rent” via speech




134

Figure 19. Petri Net showing all possible variations of data entry for month, day and time using speech, keyboard and touch screen when the “Rent” menu is selected.




135

Figure 20. Petri Net showing all operations involving various modalities as the user chooses and enters specimen data for “Start Month”, “Start Day” and “Start Time”.

5. CONCLUSION AND FUTURE WORKS

Even as we look at the current state of the art involving multimodal fusion, we realize that the multimodalities involved are pre-defined from the very start. This set-up is correct only on the condition that the fusion is implemented in a controlled environment, one in which the environment parameters remain fixed. In a real-time and real-life set-up, however, this setting is incorrect since too many parameters may change while an action (web service) is being undertaken. In this paper, we present a more flexible approach in which the user chooses the modalities that he sees fit to his situation, therefore, the fusion process is not based on the modalities that are already pre-defined from the very beginning.

In this paper, we present our approach on multimodal fusion based on the modalities that the user himself selects. The intended application is to access web services. We showed that an event involving a multimodal action is captured in an XML file clearly identifying the involved modality and its associated parameters. We showed the parsing mechanism as well as the parsing extractor. Then, the fusion of two or more modalities is presented in

concept. We then illustrate a concrete example on how our system works in practice via a car rental system. The sample application’s specification is presented using the Petri Net tool.

In this work, the user chooses the modalities he sees suitable to his situation. To further advance this work, we intend to consider the user’s interaction context (the overall context of the user, of his environment, and of his computing system) in determining which modalities are suitable to the user before the multimodal fusion process is undertaken. Acknowledgement

We wish to acknowledge the funds provided by the Natural Sciences and Engineering Council of Canada (NSERC) which partially support the financial needs in undertaking this research work. References Aim, T., Alfredson, J., et al. (2009). "Simulator-based

human-machine interaction design." International




136

Journal of Vehicle Systems Modelling and Testing 4(1/2): pp. 1-16.

Apple. (2010). "Apple Mac OS." from www.apple.com/macosx/.

Ballinger, K. (2003). "NET Web Services: Architecture and Implementation". Boston, MA, USA, Addison-Wesley.

Bates, B. and Sierra, K. (2003). " Head First Java: Your Brain on Java - A Learner's Guide", O'Reilly Media.

Bolt, R. (1980). "Put that there: Voice and gesture at graphics interface." Computer Graphics Journal of the association of computing and machinery 14(3): pp. 262-270.

Bonet, P., Llado, C. M., et al. (2010). "PIPE2." from http://pipe2.sourceforge.net/index.html.

Carnielli, W. and Pizzi, C. (2008). "Modalities and Multimodalities". Campinas, Brazil, Springer.

Caschera, M. C., D'Andrea, A., et al. (2009). "ME: Multimodal Environment Based on Web Services Architecture ". On the Move to Meaningful Internet Systems: OTM 2009 Workshops, Berlin (Heidelberg), Springer-Verlag.

Debevc, M., Kosec, P., et al. (2009). "Accessible multimodal Web pages with sign language translations for deaf and hard of hearing users". DEXA 2009, 20th International Workshop on Database and Expert Systems Application. Linz, Austria, IEEE: pp. 279-283.

Desmet, C., Balthazor, R., et al. (2005). "<emma>: re-forming composition with XML." Literary & Linguistic Computing 20(1): pp. 25-46.

Giuliani, M. and Knoll, A. (2008). "MultiML: A general purpose representation language for multimodal human utterances". 10th International Conference on Multimodal Interfaces. Crete, Greece, ACM: pp. 165 - 172.

ISO/IEC-15909-2. (2010). "Petri Nets." from http://www.petrinets.info/.

Johnston, M. (2009). "Building multimodal applications with EMMA". ICMI 05, International Conference on Multimodal Interfaces. Cambridge, MA, USA, ACM: pp.

King, K. N. (2008). "C Programming: A Modern Approach", W.W. Norton & Company.

Kress, G. (2010). "Multimodality: Exploring Contemporary Methods of Communication". London, UK, Taylor & Francis Ltd.

Lai, J., Mitchell, S., et al. (2007). "Examining modality usage in a conversational multimodal application for mobile e-mail access." International Journal of Speech Technology 10(1): pp. 17-30.

Lalanne, D., Nigay, L., et al. (2009). " Fusion Engines for Multimodal Input: A Survey". ACM International Conference on Multimodal Interfaces. Beijing, China: pp. 153-160.

Li, Y., Liu, Y., et al. (2007). "An exploratory study of Web services on the Internet". IEEE International Conference on Web Services. Salt Lake City, UT, USA: pp. 380-387.

Madani, L., Nigay, L., et al. (2005). "Synchronous testing of multimodal systems: an operational profile-based approach". IEEE International Symposium on Software Reliability Engineering. Chicago, IL pp. 334-344.

Malik, D. S. (2010). "C++ Programming: From Problem Analysis to Program Design", Course Technology.

Microsoft. (2010). "Microsoft Windows ", from www.microsoft.com/windows.

Mohan, C. K., Dhananjaya, N., et al. (2008). "Video shot segmentation using late fusion technique". 7th International Conference on Machine Learning and Applications. San Diego, CA, USA, IEEE: pp. 267-270.

Oviatt, S. (2002). "Multimodal Interfaces: Handbook of Human-Computer Interaction". New Jersey, USA, Lawrence Erbaum.

Oviatt, S., Cohen, P., et al. (2000). "Designing the user interface for multimodal speech and pen-based gesture applications: state-of-the-art systems and future research directions." Human-Computer Interaction 15(4): pp. 263-322.

Oviatt, S. L. and Cohen, P. R. (2000). "Multimodal Interfaces that Process What Comes Naturally." Communications of the ACM 43(3): pp. 45 - 53.

Pérez, G., Amores, G., et al. (2005). "Two strategies for multimodal fusion". ICMI'05 Workshop on Multimodal Interaction for the Visualisation and Exploration of Scientific Data. Trento, Italy, ACM: pp.

Petri-Nets-Steering-Committee. (2010). "Petri Nets World." from http://www.informatik.uni-hamburg.de/TGI/PetriNets/.

Pfleger, N. (2004). "Context Based Multimodal Fusion". ICMI 04. Pennsylvannia, USA, ACM: pp. 265 - 272.

PostgreSQL. (2010). from http://www.postgresql.org/. Raisamo, R., Hippula, A., et al. (2006). "Testing usability

of multimodal applications with visually impaired children." IEE, Institute of Electrical and Electronics Engineers Computer Society 13(3): pp. 70-76.

Red_Hat_Enterprise. (2010). "Linux." from www.redhat.com.

Ringland, S. P. A. and Scahill, F. J. (2003). "Multimodality - The future of the wireless user interface." BT Technology Journal 21(3): pp. 181-191.

Schroeter, J., Ostermann, J., et al. (2000). "Multimodal Speech Synthesis", New York, NY.

Sears, A. and Jacko, J. A. (2007). "Handbook for Human Computer Interaction", CRC Press.

http://www.apple.com/macosx/

http://pipe2.sourceforge.net/index.html

http://www.petrinets.info/

http://www.microsoft.com/windows

http://www.informatik.uni-hamburg.de/TGI/PetriNets/

http://www.informatik.uni-hamburg.de/TGI/PetriNets/

http://www.postgresql.org/

http://www.redhat.com/




137

Shin, B.-S., Ahn, H., et al. (2006). "Wearable multimodal interface for helping visually handicapped persons". 16th international conference on artificial reality and telexistence. Hangzhou, China, LNCS vol. 4282: pp. 989-988.

Snoek, C. G. M., Worring, M., et al. (2005). "Early versus late fusion in semantic video analysis". 13th annual ACM international conference on Multimedia. Hilton, Singapore, ACM: pp.

Steele, R., Khankan, K., et al. (2005). "Mobile Web Services Discovery and Invocation Through Auto-Generation of Abstract Multimodal Interface". ITCC 2005 International conference on Information Technology: Coding and Computing, Las Vegas, NV.

Ventola, E., Charles, C., et al. (2004). "Perspectives on Multimodality". Amsterdam, the Netherlands, John Benjamins Publishing Co.

W3C. (2010). "EMMA." from http://www.w3.org/TR/emma.

Wang, D., Zhang, J., et al. (2006). A Multimodal Fusion Framework for Children’s Storytelling Systems.

Book "LNCS ". Berlin / Heidelberg, Springer-Verlag. 3942/2006: pp. 585-588.

Wang, F., Li, J., et al. (2007). "A space efficient XML DOM parser." Data & Knowledge Engineering 60(1): pp. 185-207.

Weaver, J. L. and Mukhar, K. (2004). " Beginning J2EE 1.4: From Novice to Professional", Apress.

Wöllmer, M., Al-Hames, M., et al. (2009). "A multidimensional dynamic time warping algorithm for efficient multimodal fusion of asynchronous data streams." Neurocomputing 73(1-3): pp. 366-380.

Yuen, P. C., Tang, Y. Y., et al. (2002). "Multimodal Interface for Human-Machine Communication". Singapore, World Scientific Publishing Co., Pte. Ltd.

Zhang, Q., Imamiya, A., et al. (2004). "A gaze and speech multimodal interface", Hachioji, Japan, Institute of Electrical and Electronics Engineers Inc., Piscataway, USA.

http://www.w3.org/TR/emma

journal of emerging trends in computing and information sciences · 2011-12-20 · to assist...

Documents