decision support and business intelligence systems...
TRANSCRIPT
Decision Support and Business Intelligence Systems
(9th Ed., Prentice Hall)
Chapter 7: Text and Web Mining
Copyright © 2011 Pearson Education, Inc. Publishing as Prentice Hall 7-2
Learning Objectives
n Describe text mining and understand the need for text mining
n Differentiate between text mining, Web mining and data mining
n Understand the different application areas for text mining
n Know the process of carrying out a text mining project
n Understand the different methods to introduce structure to text-based data
Copyright © 2011 Pearson Education, Inc. Publishing as Prentice Hall 7-3
Learning Objectives
n Describe Web mining, its objectives, and its benefits
n Understand the three different branches of Web mining n Web content mining n Web structure mining n Web usage mining
n Understand the applications of these three mining paradigms
Copyright © 2011 Pearson Education, Inc. Publishing as Prentice Hall 7-4
Opening Vignette:
“Mining Text for Security and Counterterrorism”
n What is MITRE? n Problem description n Proposed solution n Results n Answer and discuss the case questions
Copyright © 2011 Pearson Education, Inc. Publishing as Prentice Hall 7-5
Opening Vignette: Mining Text For Security…
(L) Kampala
(L) Uganda
(P) Yoweri Museveni
(L) Sudan
(L) Khartoum
(L) Southern Sudan
(P) Timothy McVeigh
(P) Oklahoma City
(P) Terry Nichols
(E) election
(P) Norodom Ranariddh
(P) Norodom Sihanouk
(L) Bangkok
(L) Cambodia
(L) Phnom Penh
(L) Thailand
(P) Hun Sen
(O) Khmer Rouge
(P) Pol Pot
Cluster 1 Cluster 2 Cluster 3
Copyright © 2011 Pearson Education, Inc. Publishing as Prentice Hall 7-6
Text Mining Concepts n 85-90 percent of all corporate data is in some
kind of unstructured form (e.g., text) n Unstructured corporate data is doubling in
size every 18 months n Tapping into these information sources is not
an option, but a need to stay competitive n Answer: text mining
n A semi-automated process of extracting knowledge from unstructured data sources
n a.k.a. text data mining or knowledge discovery in textual databases
Copyright © 2011 Pearson Education, Inc. Publishing as Prentice Hall 7-7
Data Mining versus Text Mining
n Both seek for novel and useful patterns n Both are semi-automated processes n Difference is the nature of the data:
n Structured versus unstructured data n Structured data: in databases n Unstructured data: Word documents, PDF
files, text excerpts, XML files, and so on
n Text mining – first, impose structure to the data, then mine the structured data
Copyright © 2011 Pearson Education, Inc. Publishing as Prentice Hall 7-8
Text Mining Concepts n Benefits of text mining are obvious especially
in text-rich data environments n e.g., law (court orders), academic research
(research articles), finance (quarterly reports), medicine (discharge summaries), biology (molecular interactions), technology (patent files), marketing (customer comments), etc.
n Electronic communization records (e.g., Email) n Spam filtering n Email prioritization and categorization n Automatic response generation
Copyright © 2011 Pearson Education, Inc. Publishing as Prentice Hall 7-9
Text Mining Application Area
n Information extraction n Topic tracking n Summarization n Categorization n Clustering n Concept linking n Question answering
Copyright © 2011 Pearson Education, Inc. Publishing as Prentice Hall 7-10
Text Mining Terminology
n Unstructured or semistructured data n Corpus (and corpora) n Terms n Concepts n Stemming n Stop words (and include words) n Synonyms (and polysemes) n Tokenizing
Copyright © 2011 Pearson Education, Inc. Publishing as Prentice Hall 7-11
Text Mining Terminology
n Term dictionary n Word frequency n Part-of-speech tagging n Morphology n Term-by-document matrix
n Occurrence matrix
n Singular value decomposition n Latent semantic indexing
Copyright © 2011 Pearson Education, Inc. Publishing as Prentice Hall 7-12
Text Mining for Patent Analysis (see Applications Case 7.2)
n What is a patent? n “exclusive rights granted by a country to
an inventor for a limited period of time in exchange for a disclosure of an invention”
n How do we do patent analysis (PA)? n Why do we need to do PA?
n What are the benefits? n What are the challenges?
n How does text mining help in PA?
Copyright © 2011 Pearson Education, Inc. Publishing as Prentice Hall 7-13
Natural Language Processing (NLP)
n Structuring a collection of text n Old approach: bag-of-words n New approach: natural language processing
n NLP is … n a very important concept in text mining n a subfield of artificial intelligence and computational
linguistics n the studies of "understanding" the natural human
language
n Syntax versus semantics based text mining
Copyright © 2011 Pearson Education, Inc. Publishing as Prentice Hall 7-14
Natural Language Processing (NLP)
n What is “Understanding” ? n Human understands, what about computers? n Natural language is vague, context driven n True understanding requires extensive knowledge
of a topic
n Can/will computers ever understand natural language the same/accurate way we do?
Copyright © 2011 Pearson Education, Inc. Publishing as Prentice Hall 7-15
Natural Language Processing (NLP)
n Challenges in NLP n Part-of-speech tagging n Text segmentation n Word sense disambiguation n Syntax ambiguity n Imperfect or irregular input n Speech acts
n Dream of AI community n to have algorithms that are capable of automatically
reading and obtaining knowledge from text
Copyright © 2011 Pearson Education, Inc. Publishing as Prentice Hall 7-16
Natural Language Processing (NLP)
n WordNet n A laboriously hand-coded database of English
words, their definitions, sets of synonyms, and various semantic relations between synonym sets
n A major resource for NLP n Need automation to be completed
n Sentiment Analysis n A technique used to detect favorable and
unfavorable opinions toward specific products and services
n See Application Case 7.3 for a CRM application
Copyright © 2011 Pearson Education, Inc. Publishing as Prentice Hall 7-17
NLP Task Categories
n Information retrieval n Information extraction n Named-entity recognition n Question answering n Automatic summarization n Natural language generation and understanding n Machine translation n Foreign language reading and writing n Speech recognition n Text proofing n Optical character recognition
Copyright © 2011 Pearson Education, Inc. Publishing as Prentice Hall 7-18
Text Mining Applications
n Marketing applications n Enables better CRM
n Security applications n ECHELON, OASIS n Deception detection (…)
n Medicine and biology n Literature-based gene identification (…)
n Academic applications n Research stream analysis
Copyright © 2011 Pearson Education, Inc. Publishing as Prentice Hall 7-19
Text Mining Applications
n Application Case 7.4: Mining for Lies n Deception detection
n A difficult problem n If detection is limited to only text, then the
problem is even more difficult
n The study n analyzed text based testimonies of person
of interests at military bases n used only text-based features (cues)
Copyright © 2011 Pearson Education, Inc. Publishing as Prentice Hall 7-20
Text Mining Applications
n Application Case 7.4: Mining for Lies
Statements Transcribed for
Processing
Text Processing Software Identified Cues in Statements
Statements Labeled as Truthful or Deceptive By Law Enforcement
Text Processing Software Generated
Quantified Cues
Classification Models Trained and Tested on
Quantified Cues
Cues Extracted & Selected
Copyright © 2011 Pearson Education, Inc. Publishing as Prentice Hall 7-21
Text Mining Applications
n Application Case 7.4: Mining for Lies Category Example Cues
Quantity Verb count, noun-phrase count, ...
Complexity Avg. no of clauses, sentence length, …
Uncertainty Modifiers, modal verbs, ...
Nonimmediacy Passive voice, objectification, ...
Expressivity Emotiveness
Diversity Lexical diversity, redundancy, ...
Informality Typographical error ratio
Specificity Spatiotemporal, perceptual information …
Affect Positive affect, negative affect, etc.
Copyright © 2011 Pearson Education, Inc. Publishing as Prentice Hall 7-22
Text Mining Applications
n Application Case 7.4: Mining for Lies n 371 usable statements are generated n 31 features are used n Different feature selection methods used n 10-fold cross validation is used n Results (overall % accuracy)
n Logistic regression 67.28 n Decision trees 71.60 n Neural networks 73.46
Copyright © 2011 Pearson Education, Inc. Publishing as Prentice Hall 7-23
Text Mining Applications (gene/protein interaction identification)
Gen
e/P
rote
in 596 12043 24224 281020 42722 397276
D007962
D 016923
D 001773
D019254 D044465 D001769 D002477 D003643 D016158
185 8 51112 9 23017 27 5874 2791 8952 1623 5632 17 8252 8 2523
NN IN NN IN VBZ IN JJ JJ NN NN NN CC NN IN NN
NP PP NP NP PP NP NP PP NP
Ont
olog
yW
ord
PO
SS
hallo
w
Par
se
...expression of Bcl-2 is correlated with insufficient white blood cell death and activation of p53.
Copyright © 2011 Pearson Education, Inc. Publishing as Prentice Hall 7-24
Text Mining Process
Extract knowledge from available data sources
A0
Unstructured data (text)
Structured data (databases)
Context-specific knowledge
Software/hardware limitationsPrivacy issues
Tools and techniquesDomain expertise
Linguistic limitations
Context diagram for the text mining
process
Copyright © 2011 Pearson Education, Inc. Publishing as Prentice Hall 7-25
Text Mining Process
Establish the Corpus:Collect & Organize the
Domain Specific Unstructured Data
Create the Term-Document Matrix:Introduce Structure
to the Corpus
Extract Knowledge:Discover Novel
Patterns from the T-D Matrix
The inputs to the process includes a variety of relevant unstructured (and semi-structured) data sources such as text, XML, HTML, etc.
The output of the Task 1 is a collection of documents in some digitized format for computer processing
The output of the Task 2 is a flat file called term-document matrix where the cells are populated with the term frequencies
The output of Task 3 is a number of problem specific classification, association, clustering models and visualizations
Task 1 Task 2 Task 3
FeedbackFeedback
The three-step text mining process
Copyright © 2011 Pearson Education, Inc. Publishing as Prentice Hall 7-26
Text Mining Process
n Step 1: Establish the corpus n Collect all relevant unstructured data
(e.g., textual documents, XML files, emails, Web pages, short notes, voice recordings…)
n Digitize, standardize the collection (e.g., all in ASCII text files)
n Place the collection in a common place (e.g., in a flat file, or in a directory as separate files)
Copyright © 2011 Pearson Education, Inc. Publishing as Prentice Hall 7-27
Text Mining Process
n Step 2: Create the Term–by–Document Matrix
investment risk
project management
software engineering
development
1
SAP...
Document 1
Document 2
Document 3
Document 4
Document 5
Document 6
...
Documents
Terms
1
1
1
2
1
1
1
3
1
Copyright © 2011 Pearson Education, Inc. Publishing as Prentice Hall 7-28
Text Mining Process
n Step 2: Create the Term–by–Document Matrix (TDM), cont. n Should all terms be included?
n Stop words, include words n Synonyms, homonyms n Stemming
n What is the best representation of the indices (values in cells)?
n Row counts; binary frequencies; log frequencies; n Inverse document frequency
Copyright © 2011 Pearson Education, Inc. Publishing as Prentice Hall 7-29
Text Mining Process
n Step 2: Create the Term–by–Document Matrix (TDM), cont. n TDM is a sparse matrix. How can we reduce
the dimensionality of the TDM? n Manual - a domain expert goes through it n Eliminate terms with very few occurrences in
very few documents (?) n Transform the matrix using singular value
decomposition (SVD) n SVD is similar to principle component analysis
Copyright © 2011 Pearson Education, Inc. Publishing as Prentice Hall 7-30
Text Mining Process
n Step 3: Extract patterns/knowledge n Classification (text categorization) n Clustering (natural groupings of text)
n Improve search recall n Improve search precision n Scatter/gather n Query-specific clustering
n Association n Trend Analysis (…)
Copyright © 2011 Pearson Education, Inc. Publishing as Prentice Hall 7-31
Text Mining Application (research trend identification in literature)
n Mining the published IS literature n MIS Quarterly (MISQ) n Journal of MIS (JMIS) n Information Systems Research (ISR)
n Covers 12-year period (1994-2005) n 901 papers are included in the study n Only the paper abstracts are used n 9 clusters are generated for further analysis
Copyright © 2011 Pearson Education, Inc. Publishing as Prentice Hall 7-32
Text Mining Application (research trend identification in literature)
Journal Year Author(s) Title Vol/No Pages Keywords Abstract
MISQ 2005 A. Malhotra,S. Gosain andO. A. El Sawy
Absorptive capacity configurations in supply chains: Gearing for partner-enabled market knowledge creation
29/1 145-187 knowledge managementsupply chainabsorptive capacityinterorganizational information systemsconfiguration approaches
The need for continual value innovation is driving supply chains to evolve from a pure transactional focus to leveraging interorganizational partner ships for sharing
ISR 1999 D. Robey andM. C. Boudreau
Accounting for the contradictory organizational consequences of information technology: Theoretical directions and methodological implications
2-Oct 167-185 organizational transformationimpacts of technologyorganization theoryresearch methodologyintraorganizational powerelectronic communicationmis implementationculturesystems
Although much contemporary thought considers advanced information technologies as either determinants or enablers of radical organizational change, empirical studies have revealed inconsistent findings to support the deterministic logic implicit in such arguments. This paper reviews the contradictory
JMIS 2001 R. Aron andE. K. Clemons
Achieving the optimal balance between investment in quality and investment in self-promotion for information products
18/2 65-88 information productsinternet advertisingproduct positioningsignalingsignaling games
When producers of goods (or services) are confronted by a situation in which their offerings no longer perfectly match consumer preferences, they must determine the extent to which the advertised features of
… … … … … … … …
Copyright © 2011 Pearson Education, Inc. Publishing as Prentice Hall 7-33
Text Mining Application (research trend identification in literature)
YEAR
No
of A
rticl
es
CLUSTER: 1
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
05
101520253035
CLUSTER: 2
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
CLUSTER: 3
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
CLUSTER: 4
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
05
101520253035
CLUSTER: 5
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
CLUSTER: 6
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
CLUSTER: 7
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
05
101520253035
CLUSTER: 8
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
CLUSTER: 9
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
Copyright © 2011 Pearson Education, Inc. Publishing as Prentice Hall 7-34
Text Mining Application (research trend identification in literature)
JOURNAL
No
of A
rticl
es
CLUSTER: 1
ISR JMIS MISQ0
102030405060708090
100
CLUSTER: 2
ISR JMIS MISQ
CLUSTER: 3
ISR JMIS MISQ
CLUSTER: 4
ISR JMIS MISQ0
102030405060708090
100
CLUSTER: 5
ISR JMIS MISQ
CLUSTER: 6
ISR JMIS MISQ
CLUSTER: 7
ISR JMIS MISQ0
102030405060708090
100
CLUSTER: 8
ISR JMIS MISQ
CLUSTER: 9
ISR JMIS MISQ
Copyright © 2011 Pearson Education, Inc. Publishing as Prentice Hall 7-35
Text Mining Tools
n Commercial Software Tools n SPSS PASW Text Miner n SAS Enterprise Miner n Statistica Data Miner n ClearForest, …
n Free Software Tools n RapidMiner n GATE n Spy-EM, …
Copyright © 2011 Pearson Education, Inc. Publishing as Prentice Hall 7-36
Web Mining Overview
n Web is the largest repository of data n Data is in HTML, XML, text format n Challenges (of processing Web data)
n The Web is too big for effective data mining n The Web is too complex n The Web is too dynamic n The Web is not specific to a domain n The Web has everything
n Opportunities and challenges are great!
Copyright © 2011 Pearson Education, Inc. Publishing as Prentice Hall 7-37
Web Mining
n Web mining (or Web data mining) is the process of discovering intrinsic relationships from Web data (textual, linkage, or usage)
Web Mining
Web Structure MiningSource: the unified
resource locator (URL) links contained in the
Web pages
Web Content MiningSource: unstructured textual content of the Web pages (usually in
HTML format)
Web Usage MiningSource: the detailed description of a Web
site’s visits (sequence of clicks by sessions)
Copyright © 2011 Pearson Education, Inc. Publishing as Prentice Hall 7-38
Web Content/Structure Mining
n Mining of the textual content on the Web n Data collection via Web crawlers
n Web pages include hyperlinks n Authoritative pages n Hubs n hyperlink-induced topic search (HITS) alg
Copyright © 2011 Pearson Education, Inc. Publishing as Prentice Hall 7-39
Web Usage Mining
n Extraction of information from data generated through Web page visits and transactions… n data stored in server access logs, referrer logs,
agent logs, and client-side cookies n user characteristics and usage profiles n metadata, such as page attributes, content
attributes, and usage data
n Clickstream data n Clickstream analysis
Copyright © 2011 Pearson Education, Inc. Publishing as Prentice Hall 7-40
Web Usage Mining
n Web usage mining applications n Determine the lifetime value of clients n Design cross-marketing strategies across products. n Evaluate promotional campaigns n Target electronic ads and coupons at user groups
based on user access patterns n Predict user behavior based on previously learned
rules and users' profiles n Present dynamic information to users based on
their interests and profiles…
Copyright © 2011 Pearson Education, Inc. Publishing as Prentice Hall 7-41
Web Usage Mining (clickstream analysis)
Weblogs
Website Pre-Process Data Collecting Merging Cleaning Structuring - Identify users - Identify sessions - Identify page views - Identify visits
Extract Knowledge Usage patterns User profiles Page profiles Visit profiles Customer value
How to better the data
How to improve the Web site
How to increase the customer value
User /Customer
Copyright © 2011 Pearson Education, Inc. Publishing as Prentice Hall 7-42
Web Mining Success Stories
n Amazon.com, Ask.com, Scholastic.com, … n Website Optimization Ecosystem
Web Analytics
Voice of Customer
Customer Experience Management
Customer Interaction on the Web
Analysis of Interactions Knowledge about the Holistic View of the Customer
Copyright © 2011 Pearson Education, Inc. Publishing as Prentice Hall 7-43
Web Mining Tools Product Name URL
Angoss Knowledge WebMiner angoss.com
ClickTracks clicktracks.com
LiveStats from DeepMetrix deepmetrix.com
Megaputer WebAnalyst megaputer.com
MicroStrategy Web Traffic Analysis microstrategy.com
SAS Web Analytics sas.com
SPSS Web Mining for Clementine spss.com
WebTrends webtrends.com
XML Miner scientio.com
Copyright © 2011 Pearson Education, Inc. Publishing as Prentice Hall 7-44
End of the Chapter
n Questions / comments…
Copyright © 2011 Pearson Education, Inc. Publishing as Prentice Hall 7-45
All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic,
mechanical, photocopying, recording, or otherwise, without the prior written permission of the publisher. Printed in the United States of America.
Copyright © 2011 Pearson Education, Inc. Publishing as Prentice Hall