knowledge organisation by means of concept-process mapping web viewknowledge organisation by means...

Knowledge organisation by means of concept process mapping

Knowledge organisation by means of concept-process mapping

Mark GREGORY

ESC Rennes School of Business

§0 Preface to this working paperConceprocity – concept <-> process reciprocity – is a visual and textual language and toolset intended for capturing, expressing, communicating and co-creating models of topic areas of domain knowledge by domain experts or learners. This working paper introduces, positions, compares and justifies Conceprocity. The paper positions Conceprocity within existing spectra of approaches to data and knowledge organisation. It also introduces some of the more immediate use cases for which it has been created. Specifically it discusses how Conceprocity is to be used in my PhD research and teaching.

Readers of this document might like also to make reference to the online presentations of Conceprocity which can be found on the website www.MarkRogerGregory.net. In addition to a narrated audio presentation, these presentations include video demonstrating the actual construction of a Conceprocity model using Lucidchart. The presentations are of varying lengths, with the longest being intended only for people who intend actually to learn how to use Conceprocity in anger.

Readers should please understand that this is very much a working paper, not particularly well structured and in places still somewhat tentative. Nevertheless the author would very much appreciate your comments.

/tt/file_convert/5a7168457f8b9abb538cc242/document.docx

http://www.MarkRogerGregory.net/


Table of Contents

§0 Preface to this working paper.........................................................................................................1

§1 An introduction to Conceprocity.....................................................................................................6

1.1 Why model personal knowledge conceptually?................................................................6

1.2 Why Conceprocity is important and indeed essential to my work.....................................7

1.3 Designing your working life: learning how to get things done better................................7

1.4 Simple Conceprocity model...............................................................................................7

1.5 Positioning Conceprocity as a Knowledge Organisation Representation...........................8

§2 Conceprocity described..................................................................................................................9

2.1 An example Conceprocity model and how it has been created.........................................9

2.2 Conceprocity as a modelling language.............................................................................10

2.3 Conceprocity: Notions......................................................................................................11

2.4 Representing Conceprocity relationships........................................................................12

2.5 Why Conceprocity distinguishes concepts, procedures and principles............................13

2.6 Conceprocity relationships...............................................................................................14

2.7 Events and Logical Connectors.........................................................................................18

2.8 A summary of Conceprocity grammar rules.....................................................................21

2.9 Structuring Conceprocity maps........................................................................................22

2.10 Conceprocity Usage Profiles............................................................................................23

2.11 Learning to use Conceprocity: moving on from the beginner’s profile............................23

2.12 Conceprocity for the Right Brain......................................................................................24

2.13 Specific PhD research process as a Conceprocity concept map.......................................25

2 / 66


§3 The role of Conceprocity in the PhD research of Mark Gregory: some criticisms and the ways in which they are addressed in the research design............................................................26

3.1 Why Conceprocity is important in my PhD research........................................................26

3.2 The challenge according to David Weir............................................................................26

3.3 My response to David Weir’s challenge...........................................................................27

3.4 Renaud Macgilchrist’s challenge......................................................................................28

3.5 My response to Renaud Macgilchrist’s challenge............................................................29

3.6 Recap: why are concept maps essential to this Ph.D. research?......................................29

§4 Ways of organising personal knowledge and data........................................................................30

4.1 Systems thinking and modelling......................................................................................30

4.2 A Wikipedia introduction to Knowledge Organisation.....................................................31

4.3 Schema representation....................................................................................................31

4.4 Knowledge Representation..............................................................................................32

4.5 Personal Information Management System PIMS Data Structures..................................34

4.6 Knowledge Organisation: an LIS (library and information science) perspective..............38

§5 Positioning Conceprocity among Knowledge Organisation Systems.............................................39

5.1 Knowledge Representation (KR) as the primary dimension for classifying and comparing Knowledge Organisation Systems KOS...........................................................39

5.2 Analytics based on Conceprocity models.........................................................................43

5.3 A functional perspective: (Zeng, 2008)............................................................................43

5.4 Some further evaluative comments on concept mapping...............................................45

5.5 Usage profiles..................................................................................................................45

5.6 How and why Conceprocity differs from G-MOT.............................................................47

5.7 G-MOT strengths..............................................................................................................48

5.8 Conceprocity conceptual data structures........................................................................49

3 / 66


§6 A critical evaluation of Conceprocity and some suggestions for future work...............................51

6.1 The tentative nature of these initial conclusions: further research proposed.................51

6.2 More fundamental difficulties and objections.................................................................52

6.3 Towards an ontological evaluation of Conceprocity........................................................55

6.4 Learning by enquiry: some parallels with Checkland’s LUMAS........................................57

6.5 An application to student learning...................................................................................60

6.6 Complementary approaches to concept mapping as part of a mixed-methods research approach...........................................................................................................61

“What is the contribution of personal information management systems PIMS to the working model and personal work system of knowledge workers?”..........................................................61

1. Conceprocity concept-process maps. Conceprocity CAPRI or CAPRILOPE models are the result of conscious analysis and specific design by Conceprocity modellers..........................61

2. Leximancer “fuzzy” concept maps...............................................................................................61

References.................................................................................................................................................64

Table of FiguresFigure 1 The Working Model of a knowledge worker. Source: author.........................................................7Figure 2 A concrete model: Kat sitting in Mark's lounge..............................................................................9Figure 3 More general Conceprocity map..................................................................................................10Figure 4 Conceprocity representation of abstract notions and concrete facts...........................................12Figure 5 Conceprocity relationship syntax..................................................................................................13Figure 6 An association between two concepts.........................................................................................14Figure 7 An aggregation relationship..........................................................................................................15Figure 8 A composition relationship...........................................................................................................15Figure 9 Specialisation / generalisation relationship..................................................................................16Figure 10 Regulation relationships for actors.............................................................................................17Figure 11 Regulation relationships for principles.......................................................................................17Figure 12 Precedence relationships............................................................................................................17Figure 13 Basic sequence of events and procedures..................................................................................19

4 / 66


Figure 14 AND logical connector................................................................................................................19Figure 15 Inclusive OR logical connector....................................................................................................20Figure 16 An example of combining events, procedures and logical connectors.......................................20Figure 17 Summary of Grammar Rules.......................................................................................................22Figure 18 Concepts and procedures which are expanded elsewhere - signified by highlighted border.....22Figure 19 The PhD research process of the author represented as a Conceprocity concept <-> process map............................................................................................................................................................25Figure 20 A tentative set of types of KOS (from Rocha Souza et al., 2010, FIG 1)......................................40Figure 21 KOS Spectrum Source: (Daconta et al., 2003).............................................................................41Figure 22 Levels of ontological precision - (Guarino, 2006)........................................................................42Figure 23 KOS Spectrum from (Zeng, 2008, p.161) with suggested functional effectiveness.....................44Figure 24 Illustrative summary of some of our propositions......................................................................58Figure 25 Checkland's LUMAS model Source: (Checkland, 2000)...............................................................59Figure 26 Fuzzy concept map of the author's PhD journal produced using Leximancer.............................62

Table of TablesTable 1 Conceprocity Usage Profiles..........................................................................................................24Table 2 How the current research design addresses Delamont’s objections to auto-ethnography. Source: author.........................................................................................................................................................27Table 3 Knowledge representation according to (Hjørland and Nicolaisen, 2005) with additional commentary in italics.................................................................................................................................32Table 4 A suggested positioning of Conceprocity and other KOS in Functional Effectiveness terms.........44Table 5 Conceprocity Usage Profiles..........................................................................................................46Table 6 How Conceprocity differs from G-MOT.........................................................................................47Table 7 Conceptual data structures and their associated metadata..........................................................49Table 8 Conceptual Modelling Framework Elements (based on (Wand and Weber, 2002, p.364))...........56Table 9 Experiments planned or underway in the current research of Mark Gregory...............................63

5 / 66


§1 An introduction to ConceprocityConceprocity is an essentially pragmatic approach to the representation and organisation of explicit knowledge. It is based on the earlier work by the Canadian research centre LICEF (Paquette, 2010). Version 1.0 of Conceprocity was introduced on 09/05/2013. Conceprocity – concept <-> process reciprocity – is a visual and textual language and toolset intended for capturing, expressing, communicating and co-creating models of topic areas of domain knowledge by domain experts or learners. The modeller decides the vocabulary and constructs a Conceprocity model in accordance with what can be very simple grammar rules. If the modeller is prepared to learn a little bit more about Conceprocity, then more sophisticated semiotics are available to her as she take advantage of more advanced usage profiles.

1.1 Why model personal knowledge conceptually?We want to achieve models which are isomorphic with the situation that we are seeking to regulate or control. Why? (Conant and Ashby, 1970) tell us that:

“The design of a complex regulator often includes the making of a model of the system to be regulated. The making of such a model has hitherto been regarded as optional, as merely one of many possible ways… A theorem is presented which shows, under very broad conditions, that any regulator that is maximally both successful and simple must be isomorphic with the system being regulated… Making a model is thus necessary. The theorem has the interesting corollary that the living brain, so far as it is to be successful and efficient as a regulator for survival, must proceed, in learning, by the formation of a model (or models) of its environment.”

Each of us needs to regulate our own working lives. Modelling for action requires conceptual modelling of the personal work system (Baskerville, 2011) which embodies:

• The explicit elements of a person’s personal information

and

• Their intentions and means of enacting the knowledge which governs the derivation and use of that personal information.

We need a modelling formalism for concepts and behavioural elements – processes acting on concepts.

The author’s pragmatic choice is to introduce a new modelling language: Conceprocity. This modelling language is intended to allow novice and experienced knowledge mapping workers to create a myriad of models. The purpose of these models may vary. Sometimes a model will represent a means of

6 / 66


communication between stakeholders in a situation. Sometimes the process of modelling will increase the understanding of a complex situation and lead to better learning by stakeholders.

1.2 Why Conceprocity is important and indeed essential to my workPragmatism (following Pierce as collected by (Hartshorne et al., 1931)) leads us to suggest :

No model means no regulation and imperfect understanding.

Abductively, we conclude that we all have a personal work system and we must therefore have a model – but what is that model? Can we surface it, as individuals and as researchers? Surfacing this model is at the heart of my PhD research.

1.3 Designing your working life: learning how to get things done betterThe act of living our lives involves:

Your working model: your mental view of who you are and how you act

Your personal work system: how you act to get things done (Allen, 2003)

Your personal information management system: how you keep things found (Jones, 2007)

Your use of technology

1.4 Simple Conceprocity model

Figure 1 The Working Model of a knowledge worker. Source: author

Easy to criticise – but any model is better than the alternative, which is not none, but an inexplicit one!

An initial model is refined in use, in action

How you can design a better work system

You can choose between:

Explicit personal information system designHowever, for most of us with our current state of learning this is unlikely to be an option.

Serendipitous bricolage: (Ciborra and Jelassi, 1994)

7 / 66


Guided learning

My research largely takes the form of action research (design school) or observation (behaviour school) – see (Hevner et al., 2004). The work of (Hevner and Chatterjee, 2010; Hevner, 2010; Hevner et al., 2004) on design science in IS research, and of (Wand and Weber, 2002) on conceptual modelling and information systems have informed both this paper and the design of Conceprocity. To the extent that one potential use of Conceprocity is to conceptualise and therefore support the design of target personal information management systems, the design perspective identified by Hevner and his colleagues is perhaps sometimes appropriate. However, since we suspect that most personal information management systems are the result of serendipitous bricolage rather than the product of deliberate design, it is the more behavioural perspective identified by Hevner which is likely to be significant in the study of actual personal information management systems.

1.5 Positioning Conceprocity as a Knowledge Organisation RepresentationConcepts may be held both visually and linguistically. Indeed, some hold that there exists right brain and left brain thinking. This notion has developed from the late-1960s research of psycho-biologist Roger W Sperry (see for example the popular presentation (Sperry, 1975)), who discovered that the human brain has two very different ways of thinking corresponding more-or-less to the two separate brain hemispheres):

Right brain thinking is visual and processes information in an intuitive and simultaneous way, looking first at the whole picture then the details

Left brain thinking is verbal and processes information in an analytical and sequential way, looking first at the pieces then putting them together to get the whole

There are various approaches that have been developed to represent knowledge in ways which explicitly exploit right-brain approaches or seek to stimulate right-brain reactions. These include:

Mind Maps –Tony Buzan (Buzan and Buzan, 1996)

Concept maps – Joseph Novak and collaborators (Novak and Cañas, 2008) following David Ausubel (Ausubel, 2000, 1963)

Concept maps with typed concepts and relationships: LICEF (Paquette, 2010); (Basque, 2013)

Concept <-> Process maps: Conceprocity: Mark Gregory (www.markrogergregory.net)

8 / 66


Using both the visual and the linguistic (written and spoken language) stimulates better understanding of a situation and – perhaps later – better learning.

All these approaches are based on the active creation by individuals of specifically-crafted or designed maps of concepts and their relationships. An alternative and, we believe, complementary approach is discussed below in section 4.4 and included in Table 3.

§2 Conceprocity described2.1 An example Conceprocity model and how it has been

created Start with a simple English sentence: “The cat sat on the mat”

Give a specific instance: “The cat called Kat sat on the mat in my lounge”

A concrete Conceprocity map follows:

Figure 2 A concrete model: Kat sitting in Mark's lounge

Identify concepts, any static relationships and any activities

Create a specific and a more general model using the meta-concepts (Conceprocity calls them notions) of concept, procedure and relationship

9 / 66


Consider concrete and abstract representations

Observe, maybe discuss and then refine the resulting map

Here we choose to remove the concrete and retain the abstract elements in a conceptual model of the general situation of creatures acting in a geographical context

Figure 3 More general Conceprocity map

The model that results depends upon the viewpoint and the purpose of the modeller – what (Checkland, 1981) identified as the Weltanschauung of this important participant

A cat specialist (and a cat lover!) will take a different view from an expert in cognitive science applied to animals

But the process of dialogue and of mutual understanding can be aided by visual concept mapping and by dialogue around the models

2.2 Conceprocity as a modelling languageA modeller creates an initial model which is then refined (evolved and simplified)

Either alone by the original modeller

Or by means of co-modelling by the original modeller and other modellers who may be better “wielders” of Conceprocity

10 / 66


Or by exchange between the modeller(s) and domain experts who are not (yet) Conceprocity modellers

The aim is to create well-expressed models which enhance both the understanding of the phenomenon being modelled and the understanding and ongoing learning of the modeller(s).

It is essential to distinguish between Conceprocity as a modelling language, often used for co-modelling by persons more or less skilled in knowledge mapping; from the actual models produced in Conceprocity.

2.3 Conceprocity: NotionsPaquette’s G-MOT and Conceprocity differ from the more usual concept maps of (Novak and Cañas, 2008) by distinguishing between types (classes) of objects:

Concepts (things, ideas, etc.; these are usable and (sometimes) decidable classes of knowledge)

Procedures (the means of enacting knowledge in the form of specific activities, repeatable actions and processes – the latter being templates for repeated actions)

Principles (rules, constraints, permissions; also computer programs – viewed as concrete expressions of algorithms and an encoding by programmers of knowledge)

Actors (people, organisations, external systems)

Conceprocity goes beyond G-MOT in a number of ways. In particular, it supports the notions of:

Events: events describe changes of condition or state; they typically characterise the result of an activity and in turn trigger the next activity

Logical connectors: OR XOR AND NOT

Conceprocity can therefore be used for modelling algorithms and heuristics.

See Figure 4 Conceprocity representation of abstract notions and concrete facts.

11 / 66


Figure 4 Conceprocity representation of abstract notions and concrete facts

These typed classes (e.g. cat) or instances of objects (e.g. Kat) are related by relationships or relationship instances (links) which are themselves also typed.

We contend that greater semantic precision permits more expressive and more meaningful models.

2.4 Representing Conceprocity relationshipsConceprocity relationships very broadly follow UML class diagram conventions rather than G-MOT ones. This is because the UML conventions are more visually expressive than the letters used in G-MOT and therefore Conceprocity models can be made more semantically precise.

12 / 66


The relationship syntax is:

Symbol Meaning

Flow of control or of data

Influences, governs, directs…

Is instantiated as

Commentary concerning the diagram. This relationship is also used for links between icons, images and sketches and the concept to which they relate

Figure 5 Conceprocity relationship syntax

Conceprocity relationships are also referred to as notions. Thus notions in Conceprocity is a word which embraces all of concepts, procedures, actors, principles, events and relationships.

2.5 Why Conceprocity distinguishes concepts, procedures and principlesIn his book (Paquette, 2010) Gilbert Paquette suggests as a reason for distinguishing the notions of concepts, procedures and principles the need to address the weaknesses of existing modelling approaches – such as flowcharts and decision trees:

1. Imprecise meaning of the links between the entities that compose the model.2. The ambiguity in graphs where objects, actions on objects and statements of properties

that those objects possess are all mixed up and are not represented in a way that helps to differentiate them and uncover their relationships. Paquette suggests distinguishing classes of objects as concepts, actions on concepts as procedures and statements of properties as principles.

3. The difficulty of combining in one model objects which at a high summary level in the model need to be developed at a lower level with sub-models whose nature is not the same. Thus a principle at a high level might need to be developed as a procedural or conceptual sub-model.

4. Existing visual representation formalisms have emerged largely from the computer science and software engineering communities. Formalisms such as Entity Relationship models, modern structured systems analysis, conceptual graphs (John Sowa (Sowa, 2000, 1984) following Charles Pierce), the object modelling technique and the successor Unified Modelling Language UML are all representation approaches which have been built primarily for the analysis and architectural design of complex software systems. Even to read such diagrams and the links between them is hard, and to create such models

13 / 66


requires considerable expertise and an abstraction and conceptualisation capability which may be lacking among the more general knowledge workers whom Paquette (and I) wish to address and empower. Paquette states

“Our goal is different. We need a visual representation system that is both simple enough to be used by educational specialists and learners who are not computer scientists, yet general and powerful enough to represent the structure of knowledge and learning/working scenarios. The distinction and the integration of basic types of knowledge and links in the same language are essential… We present three major steps starting with (1) informal visual modelling for the educated layperson, to help represent interesting knowledge. We then (2) move onto semi-formal modelling to help define target competencies and activity scenarios for knowledge and competency acquisition by learners and workers. Finally (3) we present the more formal visual models (Ontologies) that can be used by software agents to ensure execution of knowledge-based processes on the semantic web.” [(Paquette, 2010, p.xiv) slightly amended for clarity.]

2.6 Conceprocity relationships2.6.1 The types of relationships in Conceprocity

Here we largely follow G-MOT.

Association: simple connection Aggregation: is-a, is-made-of independent

parts Composition: is-a, is-made-of dependent parts Specialisation / Generalisation: kind-of Regulation: controls, directs, influences… Precedence: comes-after, comes-before… Entrant-Product (Input-Output): causes, gives

rise to… Instantiation: is an example (instance) of…Grammar Rules govern the valid types of links that may join the knowledge types

2.6.2 AssociationsThe Association link (G-MOT: A) is simply an untyped connection between concepts. By untyped, we mean that the modeller either does not yet know the type of the relationship or is not yet capable of deciding its more precise type.

Figure 6 An association between two concepts

14 / 66


2.6.3 AggregationsThe Aggregation link (G-MOT: G) associates multiplicity – ordinality or cardinality – with a relationship.

Aggregation is an extension of the G-MOT model.

It is essential in data modelling in accordance with the relational model of (Codd 1970).

The Aggregation link (G-MOT: G) is a kind of association which says that one concept is part of another, together with others of the same type, so that all the parts are together a group of parts which constitute a whole concept: a part-whole relationship. Aggregation is a special type of association used to model a "whole to its parts" relationship. In basic aggregation relationships, the lifecycle of a part class is independent from the whole class's lifecycle. This is in contrast with composition, which follows.

Figure 7 An aggregation relationship

2.6.4 CompositionsThe Composition link (G-MOT: C) also connects a knowledge (object) with one of its constituents or its constitutive parts. The composition aggregation relationship is just another form of the aggregation relationship, but the child class's instance lifecycle is dependent on the parent class's instance lifecycle.

Thus a table is composed of legs and of a flat surface. Similarly, a chessboard is composed of squares, each of which may be black or white.

Figure 8 A composition relationship

2.6.5 Multiplicities, cardinality and ordinalityWe can ascribe a multiplicity to either or both ends of an association, an aggregation or a composition. This multiplicity can be:

An exact number A range of numbers, separated by two dots An arbitrary unspecified number

represented as an asterisk * Example multiplicities:

15 / 66


1 1..1 0..1 1..* 0..* 3..4 – e.g. number of legs on a stool 0..0 – this means that there is NO

relationship Cardinality and ordinality distinguished

Cardinality specifies the maximum number in relationships and ordinality specifies the absolute minimum number in relationships. When the minimum number is zero, the relationship is usually called optional and when the minimum number is one or more, the relationship is usually called mandatory.

2.6.6 Specialisation and GeneralisationOne knowledge type may be a particular case of another knowledge type: a sort-of relationship exists between the two concepts. This is termed specialisation. Conversely, reading the other way between the two concepts, we recognise that the first concept is a generalisation of the second. Thus a table is a specialisation of furniture. One type of furniture is a table; there are of course others.

Figure 9 Specialisation / generalisation relationship

2.6.7 RegulationThe regulation link exists to enable certain more complex links to be expressed:

In conjunction with CONCEPTs: Here the principle defines some constraints that must be satisfied or establishes a law or a relation between two or more concepts

In conjunction with a PROCEDURE OR ANOTHER PRINCIPLE: Here the principle controls or governs the execution of a procedure or the selection of other principles.

A regulation link can be used from a principle or actor towards another abstract knowledge which can be a concept, a procedure or another principle.

A regulation link between two concepts defines some constraints that must be satisfied or establishes a law that governs the relationship.

A regulation link from a principle to a procedure or to another principle controls or governs the execution of a procedure or the selection of other principles.

16 / 66


Figure 10 Regulation relationships for actors

Figure 11 Regulation relationships for principles

2.6.8 Precedence

Figure 12 Precedence relationships

The precedence link connects two procedures or principles, the first of which must be ended or estimated before the second begins or can be applied. For example: prepare an outline precedes write a text.

2.6.9 Entrant-ProductThis relationship, which is broadly equivalent to input/output, is used to identify concepts that are entrants to a procedure (the arrow goes from concept to procedure) or a product of a procedure (the arrow goes from the procedure to the concept). For example, outline is an entrant to write a text; text is the product of write a text.

A procedure may either be defined entirely informally, or it may have a more or less formal definition. If it is expressed sufficiently formally, for example in a programming language, it can be viewed as a function which transforms its entrants into its products.

2.6.10 Instantiation

17 / 66


This relationship connects an abstract knowledge to a concrete one. Abstract concepts have as their concrete equivalent facts. Abstract procedures have as their concrete equivalent traces. Abstract principles have as their concrete equivalent statements.

2.6.11 Labelling relationships in ConceprocityOne form of label which is commonly used in Conceprocity is that of multiplicity.

Conceprocity also permits relationships to be labelled with text but it doesn’t insist that that be the case. In fact the most common use of labels in relationships is in the simple usage profile of Conceprocity where the only relationship type which is supported is Association.

There is a basic issue with labelling relationships, which is that it is often necessary to label a relationship twice - once in one sense and once in another.

For this and for other reasons, Conceprocity tends to prefer the use of more notions – often procedures – to make clearer the relationship between, for example, two concepts. An intermediate procedure can be a function which transforms one concept to another. Then if concepts, procedures and principles are well named, there is no additional value in labelling relationships.

2.7 Events and Logical ConnectorsConceprocity goes beyond G-MOT in recognising the notions of event and logical connector. An event is a named change of state, typically represented in English using a past participle such as arrived or departed. Typically, events are the result of one procedure and trigger another; this is the case for P1 in the diagram which follows.

Figure 13 Basic sequence of events and procedures

18 / 66


In Conceprocity, events must be preceded and followed by either a procedure or a logical connector. The logical connectors available in Conceprocity are Exclusive-Or, Inclusive-Or and And.

Figure 14 AND logical connector

Figure 15 Inclusive OR logical connector

19 / 66


Figure 16 An example of combining events, procedures and logical connectors

2.8 A summary of Conceprocity grammar rules(Rosemann and Green, 2002) discuss how Wand and Weber have developed a series of models based on the ontological theory of Mario Bunge; they named these models the Bunge-Wand-Weber BWW models. The BWW models have as their intent the formalisation of modelling techniques. Rosemann and Green identify two particular criticisms of the models, one being the understandability of the constructs within those models and the second being the difficulty in applying the models to specific modelling techniques. They propose a meta-model of the BWW constructs using a meta-language that is in effect an extended entity relationship model. Their justification for that choice is that the entity relationship model is already very familiar to many Information Systems professionals. Furthermore the extended entity relationship model is the meta-language by which (Scheer, 2000) has modelled and formalised his

20 / 66


(Paquette, 2010) and G-MOT set out certain grammar rules which together constrain and add meaning to concept process models.

Figure 17 Summary of Grammar Rules

2.9 Structuring Conceprocity maps

21 / 66


2.9.1 The need for multi-level modelsConceprocity allows a knowledge model to be developed at several levels. The principal (top level) model gives a general overview of the domain. The subordinate models add a new level of detail, more and more specific to the representation of the domain: an expansion. Each level of representation should be simplified so as to retain only the knowledge essential at that level. By this means the benefits of encapsulation – low coupling and high internal cohesion – can be realised in the context of knowledge representation. In Conceprocity, the existence of an expansion of Concepts and Procedures is signalled by a highlighted border. The notion is then linked to another page in the same model or to another model.

Figure 18 Concepts and procedures which are expanded elsewhere - signified by highlighted border

2.9.2 Structuring a modelA Level 1 diagram is potentially supported by several Level 2 diagrams. Each Level 2 diagram might be supported by several Level 3 diagrams. There is but one Main Model and a series of nested models stemming from the Main Model.

2.10 Conceprocity Usage Profiles

22 / 66


A Usage Profile is a named usage of Conceprocity. These various usage profiles require few or no extensions to the Conceprocity basic notation which is richly expressive. It is possible and desirable to start with a beginners’ profile “Simple concept mapping for beginners”, in which the only available relationship is association and no use is made of principles, and only then to move on to typed relationships and principles. In particular in this profile, we place strong emphasis on the use of sketches, icons and images.

2.11 Learning to use Conceprocity: moving on from the beginner’s profileConceprocity is cloud-based and essentially multi-user. It also provides implicit support for the notion of usage profiles. A Usage Profile is a named usage of Conceprocity.

The various usage profiles require few or no extensions to the Conceprocity basic notation which is richly expressive.

It is possible and desirable to start with a beginners’ profile “Simple concept mapping for beginners”, in which the only available relationship is association and no use is made of principles, and only then to move on to typed relationships and principles. Particularly in this profile, we place strong emphasis on the use of sketches, icons and images.

Table 1 Conceprocity Usage Profiles

Conceprocity Usage Profile

Simple concept mapping for beginnersMakes use of a deliberately restricted range of Conceprocity objects. In particular, the only relationship type supported is association. In order to give more expressiveness, this profile permits association relationships to be named.

Knowledge mappingVery general with the full range of Conceprocity objects. Relationships should not normally be named. Instead, the nature of the two notions linked by a typed relationship should normally provide full context sufficient to make the meaning of the relationship clear. Where this is not the case, Conceprocity permits commentary/notes.Typical uses include: self-observation, research design, representing knowledge as-is and as-ought, demonstrating understanding, documenting a body of knowledge, design and delivery of teaching (e.g. to act as the “advance organiser” or signposting originally suggested by (Ausubel, 2000)), learning and evaluation, representation of algorithms and of heuristics

Use case diagrams

23 / 66


Event-driven process chains

Data (entity-relationship) models

System architecture and components

Taxonomy creation and maintenanceConceprocity is not currently intended for the representation of full ontologies; it can however be used effectively to represent taxonomies.

Causal loop diagrams (Eden, 2004)

2.12 Conceprocity for the Right BrainAs currently implemented, Conceprocity is a Lucidchart application (see https://www.lucidchart.com/). Therefore Conceprocity makes it easy to include visual elements because Lucidchart does so. Beyond Conceprocity’s own symbols, we can include images and icons. The modeller can either locate these for herself, or she can use the built-in Google Images search.

Sketches – less formal diagrams – sometimes have a role, particularly in the early development or the informal presentation of a model (especially during whiteboard sessions). This is the way in which a concept process model can include and embrace rich pictures or elements of a rich picture. Rich pictures were originally introduced by Peter Checkland (Checkland, 1981); see also (Checkland and Tsouvalis, 1997). We note that the recent widespread use of tablet computers makes it much easier to create such sketches and then to incorporate them in Conceprocity models. We note too hat sketches can be created using pen and paper and then captured digitally as images: the person responsible for the sketch takes a photograph of the outcome using her smartphone or tablet.

2.13 Specific PhD research process as a Conceprocity concept map

Figure 19 shows the author’s current PhD research process:

24 / 66

https://www.lucidchart.com/


Figure 19 The PhD research process of the author represented as a Conceprocity concept <-> process map

This map is by no means the only possible conceptualisation possible of the PhD work. Furthermore, it can easily be criticised on multiple grounds. But the very fact of there being such a model enables dialogue and offers evaluative possibilities.

§3 The role of Conceprocity in the PhD research of Mark Gregory: some criticisms and the ways in which they are addressed in the research design3.1 Why Conceprocity is important in my PhD research

Conceprocity is a semi-formal visual knowledge representation language which enables and encourages the modeller to be more precise in defining, bounding and relating conceptual and procedural knowledge.

It is in effect a means to constrain and enhance natural language expression and thereby to increase the precision of the meaning which the modeller seeks to express.

25 / 66


To the extent to which two modellers can agree upon a Conceprocity model, it is also a means to establish and to verify communication of ideas and concepts.

It or something like it is essential to completing my PhD!

My use of concept maps is motivated by the following felt needs: Structuring my understanding of the published work of others. For examples, see

(elsewhere) my concept maps concerning the work of (Weick et al., 2005) on sensemaking and of (Polya, 1988) as introduced by (Macgilchrist, n.d.) on heuristics.

Planning my PhD research, which has conceptual and process elements.

The main initial Conceprocity test use case is in fact work system modelling, particularly personal work system modelling. This is because the immediate application of Conceprocity will be to help me as the research monitor, in parallel with research volunteers, to model their and my knowledge management in the action research phase of my PhD research.

3.2 The challenge according to David WeirMy director of studies, Prof. David Weir has challenged me thus:

“I see this in a way as verging on the Autoethnographic:

http://www.qualitative-research.net/index.php/fqs/article/view/1589/3095

and thus susceptible to a special kind of critique about self-based knowledge, referentialism etc. Thus:

"Efforts at self-revelation flop not because the personal voice has been used, but because it has been poorly used, leaving unscrutinized the connection, intellectual and emotional, between the observer and the observed” (Berar) Quoted in (Spry, 2001) at http://eppl604.wmwikis.net/file/view/spry.pdf

So can Conceprocity function in such a way as to avoid critiques like those of (Delamont, 2007)1? ”

3.3 My response to David Weir’s challenge(Spry, 2001, p. 710) quotes (Denzin, 1992): “autoethnography is a radical reaction to realist agendas in ethnography and sociology… which privilege the researcher over the subject, method over subject matter, and maintain commitments to outmoded conceptions of validity, truth, and generalizability” (p. 20)”. She goes on to quote Ruth Behar, working from the writings of George Devereux as she asserts:

1 See http://www.leeds.ac.uk/educol/documents/168227.htm and http://www.cardiff.ac.uk/socsi/qualiti/QualitativeResearcher/QR_Issue4_Feb07.pdf

26 / 66

http://eppl604.wmwikis.net/file/view/spry.pdf

http://www.qualitative-research.net/index.php/fqs/article/view/1589/3095


“What happens within the observer must be made known, Devereux insisted, if the nature of what has been observed is to be understood” (Behar, 1997, p.6).

We can summarise (1) Sara Delamont’s arguments (Delamont, 2007) against auto-ethnography, which she views as pernicious and lazy and (2) the extent to which the use of Conceprocity and the overall research design might help as in Table 2 which follows:

Table 2 How the current research design addresses Delamont’s objections to auto-ethnography. Source: author

Objection to auto-ethnography Response1. Auto-Ethnography cannot fight familiarity

– it is hard to fight familiarity in our own society anyway even when we have data.

We prefer the use of the phrase “structured self-observation” (Rodriguez and Ryave, 2002) to auto-ethnography. We do this to emphasise the fact that we are not here talking about self-revelatory writing as associated with the work particularly of Carolyn Ellis or illustrated by that of Ruth Behar. In addition, the results of this structured self-observation are in no way regarded as true in an ontological sense. Instead, they are intended as a particular exemplar. Any very tentative findings or suggestions of ways forward find their way into the working documents associated with this research. Those working documents are intended to be refined in action research in which the experience of a number of research volunteers are individually expressed and then, perhaps, to a certain degree synthesised. Conceprocity gives shape and a degree of objectivity to these individual and collective expressions.

2. Auto - Ethnography is almost impossible to write and publish ethically.

3. Research is supposed to be analytic not merely experiential. Autoethnography is all experience, and is noticeably lacking in analytic outcome.

The construction of a Conceprocity model is a significant mental exercise which is inherently and intensely analytical. Although this does not in and of itself prevent an experiential and self-centred approach, it makes it less likely. Subsequently the model can be refined in discussion between an individual and a peer or mentor, and the outcome is at the very least a surfaced and carefully-expressed model.

4. It focuses on the wrong side of the power divide.

Concerning autoethnography or structured self-observation in personal information management, I genuinely doubt that any power-divide issues arise.

In connection with action research, there is

27 / 66


indeed a very serious danger associated with the unbalanced power relationship between research volunteer and research mentor and the prejudices of the latter. However, the research design emphasises and seeks to promote the establishment of a forum-based community of practice in which the voice of the research mentor will be but one of many.

5. It abrogates our duty to go out and collect data: we are not paid generous salaries to sit in our offices obsessing about ourselves. Sociology is an empirical discipline and we are supposed to study the social.

In this research design, structured self-observation is but one of a number of complementary data-collection techniques. The others are textual analysis of the existing corpus of literature concerning personal information management systems, and explicitly-authorised analysis of the writings of research volunteers. We make absolutely no pretence to statistical validity. Instead we are particularly concerned in this research to look at certain outliers - see (Boisot and McKelvey, 2010). These outliers are experts in personal information management systems who may even have a vanguard role to play in the establishment of this small field of academic enquiry.

6. Finally and most importantly ‘we’ are not interesting enough to write about in journals, to teach about, to expect attention from others. We are not interesting enough to be the subject matter.

Certainly. There is no danger that anyone’s individual experience will find its way into published work.

3.4 Renaud Macgilchrist’s challengeAccording to my colleague Renaud Macgilchrist, I need clearly to distinguish between

1. Models as pseudo-formal languages systems – e.g. predicate logic, semantic nets, frames, objects, UML and the like – even Prolog;

and:2. The infinite range of specific models, instances of models, which can be constructed using

those language systems.

I need to set out a Table of Criteria for choosing a language system which gives a weighted evaluation of the various alternative language systems.

3.5 My response to Renaud Macgilchrist’s challengeThe need to distinguish between modelling language systems and their use in expressing specific models is addressed in Table 7.

An initial table of criteria for choosing a language system appears as Table 4.

The remainder of this paper can be regarded as my more general response to this challenge.

28 / 66


3.6 Recap: why are concept maps essential to this Ph.D. research?Concepts and their relationships have been recognised as central to knowledge and understanding since the days of Plato and Aristotle. More recently, as highlighted by (Hjørland, 2009, p.1519):

“Thomas Kuhn (1922–1996) … developed a theory of concepts that corresponds with his theory of paradigms and that has been considered an important contribution to concept theory. This connection between “paradigms” and “concepts” is the point of departure for the present article. An important view of concepts today can be said to be “post-Kuhnian” in the sense that it is recognized that different theories and “paradigms” may be considered the most important mechanism for the development of concepts. However… different “paradigms” do not totally replace each other but exist together and compete with each other in all domains all the time (see, e.g., Mayr, 1997, pp. 98–994). These criticisms are the reason for using the term “post-Kuhnian” rather than “Kuhnian” in the present article. The term “post- Kuhnian” should not, however, be seen as an indication that the underlying view is primarily taken from Kuhn. There are perspectives, such as pragmatism, activity theory, and hermeneutics that are both older and have played a greater role for the views developed in the present article.”

My own stance here is both pragmatic and pragmatist. I can illustrate the pragmatic by drawing an analogy between my work here on personal information management systems within the context of personal knowledge work and that of librarians and information scientists in cataloguing and classifying knowledge made explicit as books. It is no coincidence that the majority of serious academic writing about concept mapping is associated with library and information science (but also with cognitive science). Writing a Ph.D. about personal information management is an exercise in personal knowledge management. Much of social science concerns itself with constructs and their relationships. In areas of enquiry such as that of personal information management systems - where I hold that the understanding of the field of enquiry is as yet too imprecise and incomplete to admit of the possibility of traditional positivist research - what is instead necessary is to identify and to map out concepts. Thus pragmatically I need to carry out concept mapping. The prototype pragmatist, Charles Peirce, created what he called existential graphs and what (Sowa, 1992) has more recently renamed conceptual graphs. A conceptual graph (CG) is a graph representation for logic based on the semantic networks of artificial intelligence and existential graphs. Conceptual graphs are admirably precise - they can be directly transposed into the RDF

29 / 66


semantic Web knowledge representation. I prefer concept maps because it is possible to start from the informal stance adopted by people who are not specialists in logic or computer science and then gradually, often by means of dialogue or even by dialogic mentoring, gradually to refine what is understood into ever more precise knowledge maps. These too can be formalised and directly transposed into RDF and OWL if that is appropriate. It is not appropriate when the primary purpose of concept maps is to attempt to give greater precision to the sometimes essentially imprecise or ambiguous notions partially and incompletely understood by individual knowledge workers. Thus pragmatically I have preferred Conceprocity concept and process maps to more formal knowledge representation techniques.

§4 Ways of organising personal knowledge and data4.1 Systems thinking and modelling

(Stowell and Welch, 2012, p. xiv) identify as the basic building blocks of systems thinking (1) emergence, (2) hierarchy, (3) communication and (4) control. They discuss how a system is defined from the perspective of an observer, who chooses to draw a boundary reflecting a field of interest and giving to the system so defined a name. They remind us of the taxonomy of three systemic models originally identified by Russell Ackoff (Ackoff et al., 1962) and they extend it with a fourth following Brian Wilson (Wilson, 1984) to yield:

1. An iconic model is a model of reality, the properties of which equate to those of the real article such that (albeit on a different scale) the model can be expected to behave in the same way as the real thing. I would give as an example of such a model the wind tunnel model of a new aircraft.

2. An analogical model is an attempt to simulate the behaviour of the original although its physical appearance is quite different to that of the original. Most simulation models fall into this category.

3. An analytic model is created from mathematical or logical relationships that are believed to lead to the behaviour of some situation of interest. Typical examples include spreadsheet models. Analytic models may subsequently provide the data for analogical models.

4. A conceptual model includes pictures or symbols which are used to represent the subjective and qualitative aspects of a situation.

(Stowell and Welch, 2012) present modelling as a kind of surrogate representation of some situation. It is in the process of forming, reforming and structuring that model that we begin to learn about the situation of interest and its similarities and differences to the situation that we are modelling. Among the dangers inherent in such modelling are that it becomes an end in and of itself. Instead a model is only

30 / 66


an abstraction of our perception of reality. As a simplification it is also often subjective.

Conceprocity enables the creating and maintenance of shared conceptual models.

4.2 A Wikipedia introduction to Knowledge OrganisationWikipedia http://en.wikipedia.org/wiki/Knowledge_organization accessed 12/06/2013 suggests:

“The term knowledge organization (KO) (or "organization of knowledge", "organization of information" or "information organization") designates a field of study related to Library and Information Science (LIS). In this meaning, KO is about activities such as document description, indexing and classification performed in libraries, databases, archives etc. These activities are done by librarians, archivists, subject specialists as well as by computer algorithms. KO as a field of study is concerned with the nature and quality of such knowledge organizing processes (KOP) as well as the knowledge organizing systems (KOS) used to organize documents, document representations and concepts.

The leading journal in this field is Knowledge Organization published by the International Society for Knowledge Organization (ISKO).

Simple Knowledge Organization System (SKOS) is a W3C recommendation designed for representation of thesauri, classification schemes, taxonomies, subject-heading systems, or any other type of structured controlled vocabulary. SKOS is part of the Semantic Web family of standards built upon RDF and RDFS, and its main objective is to enable easy publication and use of such vocabularies as linked data. See http://en.wikipedia.org/wiki/Simple_Knowledge_Organization_System.”

4.3 Schema representation(Paquette, 2010) discusses the relationship between structured knowledge representation and learning, which he sees as being inextricably linked. Thus understanding is impossible without identifying and classifying objects and ideas and linking them by association in some organised way. These mental structures or schemas vary in complexity. The concept of schema as the building block of mental structures is now well established in cognitive psychology. The language and the thinking derive initially from the work of Jean Piaget (Inhelder and Piaget, 1955), who discussed the meta-concepts of schema, structure, strategy and operation to describe cognitive processes. According to Piaget, growth of the intellect is achieved through increasingly logical, numerous and complex schemas. Such schemas play a central role in the construction of knowledge which in turn is essential to the learning process.

31 / 66

http://en.wikipedia.org/wiki/Simple_Knowledge_Organization_System

http://en.wikipedia.org/wiki/Knowledge_organization%20accessed%2012/06/2013

http://en.wikipedia.org/wiki/Knowledge_organization%20accessed%2012/06/2013


“Learning is a process by which a representation of a certain knowledge representation is transformed into another representation of that knowledge. Learning is a process, whereas the representation of knowledge is both the starting point and result.” (Paquette, 2010, p. 15)

The G-MOT (and therefore Conceprocity) representation system is based on the theory of schemas. We distinguish between two broad categories of schemas, these being declarative or conceptual; and procedural. The first category involves data while the second includes the procedures and methods used in processing data in order to organise information. We also follow Paquette in recognising a third category of conditional or strategic schemas which consist of principles having one or more conditions that describe context and conditional sequences. Those conditions can either be embedded in principles (in both G-MOT and Conceprocity) or they can be made explicit in the form of logical connectors attached to events (Conceprocity only).

4.4 Knowledge Representation(Hjørland and Nicolaisen, 2005) discuss knowledge representation. They remind us that “Knowledge representation is thus depending both on the objective pole: what knowledge exists to be represented and on the subjective pole: the representator or selector.” (Hjørland and Nicolaisen, 2005).

We can summarise their findings as in Table 3:

Table 3 Knowledge representation according to (Hjørland and Nicolaisen, 2005) with additional commentary in italics

Framework Technique CharacteristicsAI: symbol representation and manipulation

Logic based representations Declarative sentences and inferencing. Comment: We would suggest that propositional calculus, predicate calculus, first order logic and Horn clauses (as used in Prolog) fall within this category.

AI: symbol representation and manipulation

Procedure based representations

The meaning of a knowledge base is in its use.

AI: symbol representation and manipulation

Frame based representations “Frame-based systems are knowledge representation systems that use frames, a notion originally introduced by (Minsky, 1975) as their primary means to represent domain knowledge. A frame is a structure for representing a concept or situation such as "restaurant" or "being in a restaurant". Attached to a frame are several kinds of information, for instance, definitional and descriptive information and how to use the frame. Frames are supposed to capture the essence of concepts or stereotypical situations, for example going out for dinner, by clustering all relevant information for these situations together. This means, in particular, that a great deal of procedurally expressed knowledge should be part of the frames. Collections of such frames are to be organized in

32 / 66


Framework Technique Characteristicsframe systems in which the frames are interconnected.” (Hjørland and Nicolaisen, 2005)

AI: artificial neural networks

Parallels are drawn between neural nets and behaviourism. There is an emphasis on noting stimulus and response in an empiricist tradition and comparatively little interest in what is happening within the black box. Feedback and/or feedforward are emphasised.

Statistical analysis of large corpora of data

“The statistical approach to AI involves taking very large corpora of data, and analyzing them in great depth using statistical techniques. These statistics can then be used to guide new tasks. The resulting data, as compared to the knowledge-based approach, are extremely shallow in terms of their semantic content, since the categories extracted must be easily derived from the data, but they can be immensely detailed and precise in terms of statistical relations. Moreover, techniques - such as maximum entropy analysis - exist that allow a collection of statistical indicators, each individually quite weak, to be combined effectively into strong collective evidence. From the point of view of knowledge representation, the most interesting data corpora are online libraries of text. Libraries of pure text exist online containing billions of words; libraries of extensively annotated texts exist containing hundreds of thousands to millions of words, depending on the type of annotation. Now, in 2001, statistical methods of natural language analysis are, in general, comparable in quality to carefully hand-crafted natural language analyzers; however, they can be created for a new language or a new domain at a small fraction of the cost in human labor”

(Davis, 2001, p.8133)Large corpora of data may be approached by methods related to empiricism, which seems to be what Ernest Davis is suggesting. There is an important difference, however, between traditional empiricist approaches to knowledge representation and “text corpora” approaches. The traditional approach represents what is considered knowledge by the person doing the representation. There is only one voice present. In large corpora of texts many voices are present (what kind of voices varies according to how the text corpus is selected, e.g. if it consists of newspapers or scholarly papers).Author’s comment: textual analysis tools such as Leximancer are capable of analysing large text corpora and summarising their findings in the form of concept maps. The remark concerning “many voices” is valid and important. For this reason it is pragmatically desirable to subset large text corpora and to analyse them separately as well as together.

Semantic networks Involve nodes and links between nodes. The nodes

33 / 66


Framework Technique Characteristicsrepresent objects or contents.

(Davis et al., 1993) discuss knowledge representation. Randall Davis and his co-authors make a clear distinction between what they call reasoning and representation, which they point out are intertwined in many knowledge representations such as Minsky’s frames. They suggest user-supplied axioms, theorems and lemmas as parts of a logic-based approach to reasoning.

“The good news here is that by remaining purposely silent on the issue of recommended inferences, logic offers both a degree of generality and the possibility of making information about recommended inferences explicit and available to be reasoned about in turn” (Davis et al., 1993, p.26)

(Davis et al., 1993) do considerable service, inter alia by emphasising the necessity for triggers and procedural elements in knowledge representation and they point out that these are implicit in Minsky’s frames. However, we regard as inadmissible their suggestion of logic as a programming language – as proposed by (Kowalski, 1974) - since this approach is inaccessible to the large majority of knowledge workers. Furthermore, we are making no suggestion that Conceprocity should develop in the direction of machine execution. Instead, the emphasis is very much on enabling ordinary knowledge workers, perhaps mentored or working collaboratively, to achieve real understanding and learning about the situation that they are facing.

4.5 Personal Information Management System PIMS Data StructuresWe go on to look at approaches to structuring data as a necessary precursor to knowledgeable derivation of information.

4.5.1 Standard approaches to data and information management for individuals and small virtual teamsAn earlier paper (Gregory and Norbis, 2008) presented the hypotheses that individuals working in groups should be encouraged and educated to make better use of the available tools for information management and that the tools themselves should evolve into (or be replaced by) better ways of representing information and knowledge. See also (Sauermann et al., 2005).

Our earlier paper discussed the representation of personal data, suggesting that the ways in which data is stored on a computer influence how it can subsequently be used. We therefore identified several possible, or candidate, data representation approaches and analysed the consequences of choosing them. We suggested a classification scheme for tools based primarily on their data representation. We here reproduce and enhance the suggested categories.

4.5.2 Spreadsheets

34 / 66


Spreadsheets consist of an array of cells, each of which can store a value or a formula. A formula relates the value of the current cell to other cells which can be considered as exporting their value to be used in the formula.

Spreadsheets are a very powerful combination of the nearest approach to widely available end-user computer programming so far invented; and ways of storing (more or less) structured data in which the relationship between items of data is imposed by the use of formulae.

The paper (Gregory and Norbis, 2008) introduced the idea of what it called a functional spreadsheet which simplifies and restricts the scope of spreadsheets to make them capable of formal representation. The idea was based on an insight documented in (Peyton Jones et al., 2003). In the suggested functional spreadsheet approach, rows are hierarchically organized in an outline that groups and sub-groups the data. Cells are limited to contain only values (such as text labels, dates and numbers). Column and or row headers may contain the names of functions which may be applied either to all the values in a column or row; or to all the values in a group or sub-group defined as a hierarchical outline.

4.5.3 Relational databases, in which all data is stored as relations or setsDatabases generally have a more limited remit which they fulfil with greater precision than do spreadsheets. The most widely accepted, implemented and used type of database is the so-called “relational” database (Date 2003). He suggests as an informal initial definition that

“

A relational system is one in which the data is perceived by the user as tables (and nothing but tables); and the operators at the user’s disposal (e.g. for data retrieval) are operators that generate new tables from old. For example, there will be one operator to extract a subset of the rows of a table, and another to extract a subset of the columns – and of course a row subset and a column subset of a table can both be regarded as tables themselves. The reason such systems are called ‘relational’ is that the term ‘relation’ is essentially just a mathematical term for a table.

”

4.5.4 OutlinersAn outline is a hierarchical way to display related items of text to graphically depict their relationships. Outlining is a technique which may be implemented in general office programs or in specific computer programs known as “outliners”. An outliner is a program which stores and depicts outlines: a special text editor that allows text to be structured as an outline. Outliners are typically used for computer programming, collecting or organizing ideas, tasks or even project management. Outlining is the technique widely used in programs such as Microsoft Office PowerPoint, in which the main headings of a presentation appear as separate slides and on each slide appear points and sub-points. The same technique is available in a more powerful but perhaps less widely-used form in word processing packages such as Microsoft Office Word, which supports a very useful and underused Outline mode.

35 / 66


In an outline, a data item is given meaning by being shown in its owning hierarchy. Thus a person’s surname is a component of a composite Contact object. The relative positioning of an item conveys meaning in that the label of the owner classifies or otherwise gives contextual information concerning the owned item; and the depth in the hierarchy gives some idea of the relative importance or significance of the item.

Some programs allow a data item to participate in more than one hierarchy. Thus for example an appointment for a meeting can appear in an overall agenda or calendar, but also be linked to the name of each participant in the meeting. Effectively, the same datum is classified in more than one way. To the extent that knowledge is a product of the recognition by intelligent agents of connections between information otherwise not explicitly linked, this kind of tool can be used as a mechanism for storing relatively unsophisticated knowledge.

4.5.5 Spreadsheets as a basis for databasesAn exciting new commercial approach to the construction of database-enabled websites is the STOIC platform proposed by http://sutoiku.com/. Conversation with the founder of this company, Ismael Chang Ghalimi, suggests that reflection on personal information management was part of the motivation for STOIC. STOIC turns spreadsheets into complete applications, with a cloud-based relational database and a mobile user interface. An application is created as a new spreadsheet. An object is then a new worksheet. To add a field to an object, it is necessary to add a new column. To add a record to an object, simply add a new row.

4.5.6 A possible classification of PIMS conceptual data structuresI suggest the following initial classification; this is partially corroborated by (Völkel and Haller, 2009):

Natural language and text Tables Spreadsheets and functional spreadsheets

Spreadsheets add functional programming capability to data tables

Hierarchical Outlines Relational databases Linking and multiple classification (Tagging) Graphs and graph theory Concept maps

Formally, concept maps are graphs; the objects are nodes and the relationships are edges. The topic is extensively discussed by (Friedman and Smiraglia, 2013).

Mathematical models, statistics and matrix manipulation

Here there are a number of candidate data approaches, including categorical data analysis and clustering.

36 / 66

http://sutoiku.com/


Specific PIM (Personal Information Management) programs

Semantic web and web scienceOther approaches include object oriented databases, XML documents, RDF and OWL

See (Davies et al., 2006).

Two possibilities exist when applying semantic web approaches to personal information: either specialist PIM software or services which incorporate semantic web techniques; or systems which apply semantic web techniques to pre-existing data stored on a specific computer. The latter approach is referred to as the semantic desktop (Sauermann, 2005; Sauermann et al., 2005).

Enhancing the usability and usefulness of the Web and its interconnected resources might be achieved by:

1. Servers which expose existing data systems using the RDF and SPARQL standards. Many converters to RDF exist from different applications. Relational databases are an important source. The semantic web server attaches to the existing system without affecting its operation.

2. Documents “marked up” with semantic information (an extension of the HTML <meta> tags used in today’s Web pages to supply information for Web search engines using web crawlers). This could be machine-understandable information about the human-understandable content of the document (such as the creator, title, description, etc., of the document) or it could be purely metadata representing a set of facts (such as resources and services elsewhere in the site). (Note that anything that can be identified with a Uniform Resource Identifier (URI) can be described, so the semantic web can reason about animals, people, places, ideas, etc.) Semantic mark-up is often generated automatically, rather than manually.

3. Common metadata vocabularies (ontologies) and maps between vocabularies that allow document creators to know how to mark up their documents so that agents can use the information in the supplied metadata (so that Author in the sense of ‘the Author of the page’ won’t be confused with Author in the sense of a book that is the subject of a book review).

A very important issue: whose ontology?If we accept the necessity for imposing some sort of classification mechanism to achieve accuracy and precision in searching for information, the next question which inevitably arises is “whose ontology shall we adopt?” We can identify three broad and overlapping alternatives:

Standardisation by committee (or by professional body, or by employer): top-down impositionThis is frequently done within communities of experts, such as pharmacists or medical practitioners.

Emergent ontology - ontologies shared between workers in small, often virtual, groups: bottom-up conceptualisation

37 / 66


This situation is common in areas of fast-changing technology or practice. A common vocabulary and classification system “emerges” and almost imposes itself. Evolution, when it occurs, is ad hoc.

Specialist programs which recognise or implement user-defined ontology

4.6 Knowledge Organisation: an LIS (library and information science) perspectiveSection 4.2 introduced knowledge organisation KO. In this section we establish a connection between concepts and the KO Knowledge Organization sub-domain of LIS library and information science.

(Friedman and Smiraglia, 2013) seek to

“improve comprehension of the socially-negotiated identity of concepts in the domain of knowledge organization. Because knowledge organization as a domain has as its focus the order of concepts, both from a theoretical perspective and from an applied perspective, it is important to understand how the domain itself understands the meaning of a concept. ” (Friedman and Smiraglia, 2013, p.27).

They do so within the frame of what they mean by Knowledge Organisation KO, which is identified as the:

“entire population of formal proceedings in knowledge organization – all proceedings of the International Society for Knowledge Organization’s international conferences (1990-2010) and those of the annual classification workshops of the Special Interest Group for Classification Research of the American Society for Information Science and Technology (SIG/CR).” (Friedman and Smiraglia, 2013, p.27).

Thus knowledge organization KO is defined and identified with the perspective adopted by ISKO and its journal, the Journal of Knowledge Organization. Most writers in this field come from the LIS Library and Information Science discipline. Their perspective is very important but subtly different both from the information systems discipline and the cognitive science approaches.

The literature on knowledge organisation makes frequent reference to knowledge organisation systems KOS. Although I instinctively dislike this term because the literature surrounding it does not always appear to understand what a system is, I will use it because others do.

§5 Positioning Conceprocity among Knowledge Organisation Systems

38 / 66


We have already mentioned Ernest Davis’ summary: (Davis, 2001) of knowledge representation.

5.1 Knowledge Representation (KR) as the primary dimension for classifying and comparing Knowledge Organisation Systems KOSSee (Rocha Souza et al., 2010) for an extensive discussion of knowledge organisation systems (KOS) and the related topic of knowledge representation.

They themselves use a CMap concept map (cf. Novak and Cañas, 2008) to suggest a tentative set of types of KOS (Rocha Souza et al., 2010). This is reproduced here as Figure 20:

Figure 20 A tentative set of types of KOS (from Rocha Souza et al., 2010, FIG 1)

Conceprocity could appear at more than one point in this essentially hierarchic classification, since it can be used as a form of concept mapping system, but it can also be used to make a taxonomy or classification scheme (Jacob, 2004) and in the construction of semi-formal ontologies.

39 / 66


(Rocha Souza et al., 2010: figure 3) reproduces a KOS Spectrum originally proposed by (Daconta et al., 2003). It is repeated here as Figure 21:

Figure 21 KOS Spectrum Source: (Daconta et al., 2003)

This uses a single dimension, identified as semantic strength, along which they position various KOS. (Rocha Souza et al., 2010) comment that (Daconta et al., 2003) tend to present KOS and their representational languages together; but we would comment that they themselves tend to identify and equate KR and KOS as we have previously seen in Figure 20. Here in Figure 21, travelling from weak semantics to strong is identified with the formality of the knowledge representation language.

Conceprocity has rather weak semantics. It can be used to represent data structures in a way which is as precise as the ER Entity Relationship model and indeed Chen’s E/R Entity Relationship model (Chen, 1976).

Conceprocity is in fact considerably more expressive than even an extended ER model, which concerns itself with data and its semantics. Conceprocity recognises procedures, logical connectors, principles and events. However the semantics associated with the event and process elements are deliberately not very precise

40 / 66


since they depend on the modeller’s use of natural language text. It is possible to represent relational data structures, processes, events and Boolean decisions in Conceprocity but no pretence is made to sufficiently formal semantics to permit, for example, the execution of a Conceprocity model – it is not a programming language and does not support automatic inferencing. In any concept mapping approach, the precise form of the name given to a concept and to a relationship – expressed as they are in natural language – are at one and the same time extremely important and extremely difficult to get right. We present Conceprocity as a visual language and intend it to be used for communication between human observers.

(Rocha Souza et al., 2010) quote (Guarino, 2006) as proposing the term « ontological precision ». They reproduce a figure from Nicola Guarino as their FIG. 8, reproduced here as Figure 22:

Figure 22 Levels of ontological precision - (Guarino, 2006)

(Guarino, 1998) had earlier clarified the terminology associated with ontology as a philosophical term versus ontology as an artefact in the following terms:

Ontology: the philosophical discipline Study of what there (possibly) is Study of the nature and structure of reality

Domain of entities Categories and relations

41 / 66


Characterizing properties

An ontology: a theoretical or computational artefact

“An explicit and formal specification of a conceptualization” (Gruber, 1993)

A specific artefact expressing the intended meaning of a vocabulary in terms of the nature and structure of the entities it refers to

In its current form, Conceprocity can be used to represent taxonomies and informal ontologies.

5.2 Analytics based on Conceprocity modelsConceprocity models are stored in the Lucidchart server system. They can be saved in .vdx format, this being the Microsoft Visio XML format – see http://msdn.microsoft.com/en-us/library/office/aa218409%28v=office.10%29.aspx

It is therefore potentially possible to:

1. Create concept dictionaries and a database of Conceprocity terms and usages; these in turn can form the basis for semi-automatic ontological analysis and perhaps comparison of personal ontologies.

2. Carry out various forms of data analysis, potentially including metrics such as Betweenness, Closeness, Diameter, Clustering Coefficient, Average shortest path…

See https://gephi.org/ (but no direct XML import) and http://www.liquid-technologies.com/xml-schema-editor.aspx

I do NOT anticipate carrying out such analysis in PhD timescales.

5.3 A functional perspective: (Zeng, 2008)(Zeng, 2008, p.161) suggests a table of functional effectiveness reproduced as FIG 9. in (Rocha Souza et al., 2010) and below as Figure 23:

42 / 66

http://www.liquid-technologies.com/xml-schema-editor.aspx%20

http://www.liquid-technologies.com/xml-schema-editor.aspx%20

https://gephi.org/

http://msdn.microsoft.com/en-us/library/office/aa218409(v=office.10).aspx%20


Figure 23 KOS Spectrum from (Zeng, 2008, p.161) with suggested functional effectiveness

We now suggest a positioning of Conceprocity based on Figure 23 – see Table 4 following, which:

1. reproduces Zeng’s table2. extends it with additional columns both with two additional KOS which I here identify

([1]the semantic web and its RDF and OWL knowledge representation and [2] first-order logic) and also with Conceprocity

3. extends it with additional rows: visualisation and suitability for machine processing

Table 4 A suggested positioning of Conceprocity and other KOS in Functional Effectiveness terms

Function ↓ KOS →

Term lists

Synonym rings

Gazetteers and

directories

Classification

schemes and

Taxonomies

Thesauri

Ontologies and

semantic

networks

Semantic w

eb: R

DF, O

WL

First order logic

Conceprocity

Eliminating ambiguity *** *** ** **** ** *** *** *

Controlling synonyms **** *** ** **** **

Establishing hierarchical relationships

* **** *** *** ** ** ***

Establishing associative

**** ***** ** ** ***

43 / 66


relationships

Presenting properties ***** * * **

Visualisation ** ****

Machine processing ***** ****

5.4 Some further evaluative comments on concept mapping(Trochim, 1989) introduces the use and usefulness of concept maps for planning and evaluation. The same author discusses their reliability in (Trochim, 1993). More recently he and colleagues have discussed concept mapping as an alternative approach for the analysis of open-ended survey responses (Jackson and Trochim, 2002).

More speculatively, (Thimor and Fidelman, 1995, page 35) have suggested:

“Concepts are related to logic. There are several approaches to logic and in every approach concepts are created and interrelated differently. There are two possibilities for the hierarchic set-up of hierarchic concept-maps: one possibility is a top-down map; the other is a bottom-up map. Therefore, individual differences in concept-mapping may be related to individual differences in the approach to logic. It was [earlier] suggested by Fidelman that different approaches to logic may be related to individual differences in the relative efficiencies of the hemispheric analytical and synthetical mechanisms. In the sequel a theory relating concept mapping to the cerebral hemispheres through logic is suggested. According to this theory, top-down concept-mapping is more related to the right hemisphere relative to bottom-up mapping. On the other hand, bottom-up mapping is more related to the left hemisphere relative to top-down mapping.”

They go on to suggest a contrast between the formation of concepts in Bertrand Russell’s logic which they present as bottom-up, and on the other hand, in Gottlieb Frege’s logic concepts – where a top-down formation is primary: (Thimor and Fidelman, 1995, page 50).

This suggests that concept mapping can appeal to different populations who will tend to use it in various complementary ways, particularly with the emergence of cloud-based approaches based on collaborative infrastructure.

5.5 Usage profilesConceprocity is cloud-based and essentially multi-user. It also provides implicit support for the notion of usage profiles. A usage profile is a named usage of Conceprocity.

44 / 66


The various usage profiles require few or no extensions to the Conceprocity basic notation which is richly expressive.

It is possible and desirable to start with a beginners’ profile “simple concept mapping for beginners”, in which the only available relationship is Association and no use is made of principles, and only then to move on to typed relationships and principles. Particularly in this profile, we place strong emphasis on the use of sketches, icons and images. We have chosen to give the name CAPRI to this simple usage profile. CAPRI stands for concepts, actors, procedures, relationships, images. By way of contrast, we identify the full usage profile as CAPRILOPE, Concept / Actor / Procedure / Relationship / Image / Logical Operator / Principle / Event.

Table 5 Conceprocity Usage Profiles

Model type Name PurposeSimple concept mapping for beginners

Conceprocity CAPRI Concepts Actors Procedures Relationships Images. Makes use of a deliberately restricted range of Conceprocity notions. In particular, the only relationship type supported is Association. In order to give more expressiveness, this profile permits Association relationships to be named.

Knowledge mapping

Conceprocity CAPRILOPE

Very general with the full range of Conceprocity objects, Concept / Actor / Procedure / Relationship / Image / Logical Operator / Principle / Event. In this profile, relationships should not normally be named. Instead, the nature of the two notions linked by a typed relationship should normally provide full context sufficient to make the meaning of the relationship clear. Where this is not the case, Conceprocity permits commentary / notes.Typical uses include: self-observation, research design, representing knowledge as-is and as-ought, demonstrating understanding, documenting a body of knowledge and design of teaching, learning and evaluation. In the context of teaching, it is sensible to use such knowledge maps as the “advance organiser” or signposting originally suggested by (Ausubel, 1963). This usage profile is also suitable for the representation of algorithms and of heuristics.

Use case diagrams

Conceprocity Use Case

Conceprocity use case diagrams are generally similar to UML UCDs but they are extended to show the interaction between an actor and a use case as a specific interaction element. This is done because such interactions normally need to be implemented, sometimes as form and subform hierarchies, sometimes as webpage hierarchies

45 / 66


Event-driven process chains

Conceprocity EPC Conceprocity event process chain diagrams are generally similar to ARIS EPC diagrams but they are optionally extended by incorporating a specific Data swimlane. The data swimlane is populated by concepts, which may subsequently be implemented as data tables, data views, specific file-types or by webpages. The value of the data swimlane is that interactions between it and other (non-data) swimlanes enable the modelling of the data flows (dataflows) that would otherwise require specific dataflow diagrams (DFDs)

E/R Data models Conceprocity E/R Conceprocity Entity / Relationship diagrams follow the conventions established by (Chen 1976) and subsequent work. However, ordinality, cardinality and multiplicity are shown in the Conceprocity / UML style because this is more expressive than Chen’s notation

A notion similar to that of usage profile does exist in G-MOT. One usage profile that exists in G-MOT which has no direct equivalent in Conceprocity is that of an ontology builder and the automatic generation of OWL language statements from the ontology model which is so built.

5.6 How and why Conceprocity differs from G-MOTSo why not simply reuse the existing G-MOT formalism? Table 6 gives a (gentle) critique of Mot+ and G-MOT and outlines how Conceprocity differs:

Table 6 How Conceprocity differs from G-MOT

G-MOT Conceprocity

Based on the object oriented (OO) approach extensively used in software engineering, but just as the OO approach is often vague about its philosophical and pragmatic antecedents, so (sometimes) is G-MOT

Conceprocity is a little closer to UML – particularly in the ways in which concepts are related

G-MOT is object-influenced, most obviously by class diagrams. But it separates procedures out from concepts, thus eschewing encapsulation – which has value in software engineering but not always in clarifying meaning and understanding

Conceprocity follows G-MOT. Inheritance is explicitly supported between concepts by means of a specialisation- generalisation relationship. The effect of encapsulation can be achieved by deft use of hierarchy: an apparently atomic concept is expanded at a lower level in the modelling hierarchy.

The visual representation used is sometimes obscure, specifically in the areas of how the different types of

Conceprocity prefers a UML-influenced style in which the type of arrow shows the kind of relationship.

46 / 66


G-MOT Conceprocity

relationship are displayed; they are signified by a character label rather than by a visual device

This is initially a little more difficult to teach and learn, but subsequently makes Conceprocity models easier to read and to understand.

The visual representation used is sometimes unclear, particularly the visual distinction between classes and object-instances (although this is better in G-MOT than in the earlier Mot+)

Conceprocity is clearer again in this respect.

The expression is not very visual, depending too much on textual elements and not on images and icons: It does not engage the right brain

Particularly in the simple usage profile, users are actively encouraged to make full use of icons, images and sketches.

It does not permit the clear expression of algorithms, in particular conditionality (if… then… else… endif) and repetition (do while…; repeat until…)

Whereas in G-MOT conditional statements are represented as principles, Conceprocity prefers to make this visually much clearer by using logical connectors and the separate event syntax (here following the event process chain paradigm suggested by(Scheer et al., 2005).

The language does not encourage consideration of object state and/or events

Conceprocity uses the event notion to make this much clearer.

Cardinality and ordinality (multiplicity) is not made explicit in associations

Conceprocity follows the conventions of UML class diagrams in this respect, making multiplicity much more evident.

G-MOT is a standalone (“desktop”) application available only for Windows. It is therefore not SaaS, software as a service – which is needed to make web-based collaboration on concept maps possible and easy – especially in the context of the PhD research which I am currently undertaking

Conceprocity is implemented using the Lucidchart web-based diagramming system, which is SaaS.

5.7 G-MOT strengths

47 / 66


G-MOT has many strengths, some of which are not matched in Conceprocity. Its visual editor, although restricted to the Windows desktop, enforces grammar rules much more effectively than does Conceprocity. A notion similar to that of usage profile does exist in G-MOT. One usage profile that exists in G-MOT which has no direct equivalent in Conceprocity is that of an ontology builder and the automatic generation of OWL language statements from the ontology model which is so built.

5.8 Conceprocity conceptual data structuresIn any knowledge representation scheme, it will normally be necessary also to represent data. Table 7 is an attempt at a synthetic view and positioning of Conceprocity within the spectrum of conceptual data structure CDS, here following (Völkel and Haller, 2009) :

Table 7 Conceptual data structures and their associated metadata

Technique Metadata Expressiveness, precision and recall

Spreadsheets Pragmatic – the meaning of the data is not explicit, but is partially expressed in the natural language semantics of column and/or row headings; and partially in relationships expressed as formulae between cells

Potentially very expressive and frequently imprecise or even contradictory. Charting permits visually-arresting representations of some of the underlying data.

Relational databases If the data is normalised (Codd, 1971; Date, 2003), then the column headings name sets of atomic (non-divisible) data items. This is deliberately constricting, because human-readable metadata, in the form of a natural language description (name) for each attribute, can be exploited by users as they enquire from the data, enabling precise answers to questions they have. These names can be extended by a data dictionary (which, however, is often not accessible to the end-user of the data in the database).

Deliberately very restricted expressiveness. All data is constrained to appear as tables to permit generality and precision of subsequent querying. The results of queries are themselves virtual tables constructed from the original input data.

Outlining and Outliners

The relative positioning of the items in a hierarchy groups and classifies data; and associates meaning with each group and sub-group. The addition of a grid, as in the products Ecco and InfoQube – see (Gregory, 2010) - permits further structuring and expressiveness.

Hierarchies themselves are cognitively powerful or not depending on the prior training of the user.

Mindmaps The relative positioning of the items in a diagram groups and classifies data; and associates meaning with each branch and sub-branch. An image is (potentially) associated with each branch or sub-branch

Visually very powerful, the user perceives both structure and meaning. Querying is very imprecise or non-existent.

48 / 66


Technique Metadata Expressiveness, precision and recall

Concept maps (Novak and Cañas, 2008)

The relative positioning of the items in a diagram groups and classifies data; and associates meaning with each branch and sub-branch.

Visually very powerful, the user perceives both structure and meaning. Relationships are distinguished from concepts. Querying is very imprecise or non-existent in current implementations.

XML, RDF and OWL

The meaning of an XML document is described in an associated Data Type Definition (DTD) or Schema. The RDF Schema carries this forward.

XML-based approaches potentially combine the strengths of outlining and of relational database. Because XML is both a language and a meta-language, it is possible to define specialised languages such as OPML.Generalised query languages for XML data are emerging. RDF makes possible the expression of simple forms of knowledge (as opposed simply to information), and supports processes like:

User-specified keyword classification of information structured in accordance with user design

Rule-based auto-classification Tagging

Conceprocity maps The relative positioning of the items in a diagram groups and classifies data; and associates meaning with each branch and sub-branch. An image is (potentially) associated with each branch or sub-branch. Each object has a type, as does each relationship (link). Appropriate use of hierarchy enables encapsulation.

Visually very powerful, the user perceives both structure and meaning. Querying is currently non-existent but because objects are semantically classified it would be relatively straightforward to construct a dictionary for each Conceprocity map and a lexicon (index) across multiple maps.

The latter in a cloud context might permit the emergence of shared ontologies, especially if the maps are constrained to conform to RDF and OWL standards. The latter is not yet proposed for Conceprocity but has been achieved with extensions to the similar G-MOT approach. See (Paquette, 2010: part 3)

First order logic and Horn clauses

Expressiveness and precision very high; readability and visual appeal very limited (although these can be enhanced by the use of libraries which create visualisations from Prolog statements). Querying is very general and strong logical inferencing capabilities are offered

49 / 66


§6 A critical evaluation of Conceprocity and some suggestions for future work6.1 The tentative nature of these initial conclusions:

further research proposedConceprocity is a semi-formal visual knowledge representation language which enables and encourages the modeller to be more precise in defining, bounding and relating conceptual and procedural knowledge. It is in effect a means to constrain and enhance natural language expression and thereby to increase the precision of the meaning which the modeller needs to express. To the extent to which two modellers can agree upon a Conceprocity model, it is also a means to establish and to verify communication of ideas and concepts.

In this paper in its present form I have sought to respond in a positive manner to the constructive criticism of David Weir and of Renaud Macgilchrist. In doing so I have both myself learned and also – I hope – begun to position the next phase of my research. I look forward to further criticism!

Certainly, Conceprocity is not without its weaknesses. It is arguably an error to permit so much generality of expression in a single modelling approach. The counter-argument is that usage profiles permit a more restricted representation and are therefore less likely to give rise to cognitive overload in users and readers. I would also point out that in knowledge representation schemes such as UML, it is necessary to learn a wide range of different – sometimes annoyingly so – representations. This problem is even starker in the area of conventional structured analysis, where a simple notion such as process is represented in different, overlapping and confusing ways – contrast data flow diagrams, event process chains and use case diagrams.

Because Conceprocity has been built using the Lucidchart diagramming system, it will be very easy to change the formalism in directions suggested by the readers of this article. For this and many other reasons I would strongly welcome critique and suggestions for improvement.

A potential disadvantage of moving away from the simple concept and relationship approach to concept mapping which most concept maps (and mappers) employ is to make it slightly more difficult to create metrics. Whereas (Friedman and Smiraglia, 2013) can present concept maps simply as nodes and arcs, Conceprocity maps include more basic notions – both nodes and arcs have type - and will therefore be slightly more difficult to analyse programmatically.

In the next few months I intend using Conceprocity in the following ways:

50 / 66


1. I will gradually build a Conceprocity map of my own Personal Work System PWS and the supporting Personal Information Management System PIMS. This will have two forms, an as-is – current map – and an as-ought – what I would like to achieve. These models will be based on the PhD journal which I have now maintained for two years.

2. I have already used G-MOT to analyse complex scientific articles and will in future use Conceprocity to do so.

3. Two of my masters’ research students will use Conceprocity as a small element of their research projects, due for submission in November.

4. In the first semester of the new academic year I will teach two sessions on concept mapping within my IS505E Principles of E-Commerce PEC module. I expect two classes each of about 40 students. The first session will introduce the basic usage profile of Conceprocity and the second more advanced usage. One of the two courseworks in this module requires each student to choose an academic article concerning e-commerce or information systems and to analyse its content. Students will be required to use the basic Conceprocity usage profile CAPRI and encouraged to use the full usage profile CAPRILOPE as part of the article analysis. This action learning approach will give me valuable feedback on the practical and conceptual difficulties faced by students as they learn Conceprocity. This will in turn be fed back into an improved Conceprocity 2.0 early in the new calendar year.

5. I hope to be able to carry out some quantitative assessment of Conceprocity models and modelling based on the IS505E PEC experience. This is dependent on being able to carry out some software-assisted analysis of Conceprocity models stored in XML format. This is perhaps a student project or an internship possibility for a software engineering student from a software engineering school.

Together these experiments will yield data and enable me (and a co-author?) to create a research-based journal article on concept process mapping early in 2014. This will represent a useful contribution from the PhD process.

6.2 More fundamental difficulties and objectionsWe have largely accepted as a given the notions put forward by (Paquette, 2010) which are themselves based partly on the UML thinking of (Booch et al., 2005). Paquette’s thinking also derives in large part from cognitive science; this influence pervades his book and in particular informs chapter 6 on taxonomies of problems and generic skills.

In section 2.5 above, we suggested that existing visual representation formalisms have emerged largely from the computer science and software engineering communities. It is instructive to reconsider the origins of formalisms such as Entity Relationship models, modern structured systems analysis, conceptual graphs (John Sowa (Sowa, 2000, 1984) following Charles Pierce), the object modelling technique and the successor Unified Modelling Language UML. These

51 / 66


are all representation approaches which have been built primarily for the analysis and architectural design of complex software systems. In Conceprocity as it currently stands we have designed and presented a visual representation system which, following (Paquette, 2010, p.xiv), we wish to be usable by educational specialists and learners who are not computer scientists. It is at the same time general and powerful enough to represent the structure of knowledge and learning/working scenarios.

Paquette goes on to say:

“We present three major steps starting with (1) informal visual modelling for the educated layperson, to help represent interesting knowledge. We then (2) move onto semi-formal modelling to help define target competencies and activity scenarios for knowledge and competency acquisition by learners and workers. Finally (3) we present the more formal visual models (Ontologies) that can be used by software agents to ensure execution of knowledge-based processes on the semantic web.” [(Paquette, 2010, p.xiv) slightly amended for clarity.]

Thus G-MOT supports three dialects, one for general use, one for instruction design and one for ontology building. Similarly Conceprocity distinguishes usage profiles (see Table 1) within a single visual representation language.

Recall that notion is the name given in Conceprocity to the modelling meta-concepts of concepts, procedures, actors, principles, events and relationships. See sections 2.3 and 2.4. A possible alternative word for notions is meta-concepts, that is, concepts about concepts. We now wish further to address the issue of whether Conceprocity has chosen the right notions. In section 2.5 we discussed why Conceprocity distinguishes concepts, procedures and principles. Here we consider the nature of concept mapping itself and the relationships permitted in Conceprocity.

6.2.1 What is concept mapping anyway?Much of the literature surrounding concept mapping comes from the field of enquiry known as knowledge organisation which is largely situated within the discipline known as library information science. (Hjørland, 2009) holds that information science and knowledge organization cannot avoid relating to theories of concepts. Knowledge organizing systems (e.g., classification systems, thesauri, and ontologies) should be understood as systems organizing concepts and their semantic relations. Different theories of concepts have different implications for how to construe, evaluate, and use such systems. Based on what he calls “a post-Kuhnian view” of paradigms, Hjørland argues that the best understanding and classification of theories of concepts is to view and classify them in accordance with epistemological theories (he emphasises empiricism, rationalism, historicism, and pragmatism). Different views of concepts are associated with different worldviews and epistemologies which tend to compete with each other. The historicist and pragmatist understandings of concepts are in his view the most fruitful

52 / 66


views; he outlines the importance of historicist and pragmatic theories of concepts for information science. For him, the concept is a socially negotiated construct that should be identified by studying discourses (Hjørland, 2009, p.1530). This view of concept theory has been labelled socio-constructivist.2

(Friedman and Thellefsen, 2011) discuss knowledge organisation systems and the emergence of concept theory and semiotics in that connection. For them knowledge organisation as a domain has as its focus the order of concepts, both from a theoretical perspective and from an applied perspective. It is therefore important to understand the meaning of a concept found in text and in visual maps. Whatever the epistemological stance one adopts, it is evident that the meaning of a concept is that which was intended by the originator of that concept in accordance with their own particular epistemological stance.3

Thus when (Friedman and Smiraglia, 2013) attempt a synthesis of the existing theory concerning concepts and concept mapping they do so within the tradition of library information science and in particular they identify “knowledge organisation systems”, based on earlier work reported as (Friedman and Thellefsen, 2011).

6.2.2 Relationships in ConceprocitySome concepts refer to data. The E/R Entity Relationship model of (Chen, 1976) has informed in particular Conceprocity’s thinking about associations, cardinality, ordinality and multiplicities. See section 2.6.5

The ideas of aggregation, generalisation and specialisation were introduced by (Smith and Smith, 1977) and later informed the design of UML and G-MOT. However, it is difficult to discern a single source of inspiration for the conceptualisations underlying UML. Specifically, UML does not possess a meta-model; nor does G-MOT. See section 2.6.3

Composition and part-whole relationships are the subject of mereology (which is separate from the concept of topology). (Guarino and Poli, 1995) give a fuller introduction. See section 2.6.4.

6.2.3 Conceprocity: Design issues which are not yet fully resolvedIn its current guise, Conceprocity recognises various structural relationship types. These are:

Association Aggregation Composition Specialisation / Generalisation Regulation

2 Kuhn also wrote extensively about the nature of concepts, largely from a cognitive psychology point of view.

3 I note here that the stance of Hjørland is socio-constructivist and that my own is pragmatist – a stance which Hjørland also recognises as very important.

53 / 66


Precedence Input-Product Instantiation

Are there other structural relationship types beyond those already recognised in Conceprocity?

Is there (for example) a relationship type is-a-model-of in the relationship concept-A is-a-model-of concept-B?

Given that knowledge types share common characteristics with UML classes, but that knowledge is not at all the same thing as data: What interpretations can we validly give to the relationship types which Conceprocity recognises? Thus, what is inheritance?

Note here that the semantics of inheritance are debated within the programming community. See for example: (Cook et al., 1990) who suggest that in typed object-oriented languages the subtype relation is typically based on the inheritance hierarchy. They therefore put forward a new typed model of inheritance that introduces polymorphism into the typing of inheritance. They call for the uniform application of inheritance to objects, classes and types. The resulting notion of type inheritance allows them to show that the type of an inherited object is an inherited type but not always a subtype. This more general view of inheritance is I suggest not only sensible but also what most human users would expect when reading a Conceprocity model in which one concept is a specialisation of another.

Similarly, is multiple inheritance meaningful for Conceprocity concepts? Certainly, it is possible to show it in Conceprocity.

How in Conceprocity do we best represent the subsetting and overlapping sets which are well summarised by Venn diagrams?

The G-MOT editor continues to have strengths not possessed by Conceprocity; these include XML import and (better) export, explicit support for the OWL Web Ontology Language, and copy with reference between diagram levels. The latter would ideally be generalised in Conceprocity…

Conceprocity does not yet possess easy means to produce and maintain a dictionary of the objects it contains, nor any metrics.

6.3 Towards an ontological evaluation of Conceprocity(Wand, 1996) holds that despite the availability of a large number of systems analysis methods and techniques there does not exist a general underlying foundation for this knowledge domain. The stance which Wand adopts is that an information system is a representation of another “real-world” system. This ontological stance borrows from the philosophy of Mario Bunge, and in particular his ontological formalism as presented in (Bunge, 1977, 1979). Wand sees an information system as a representation that enables us to obtain knowledge about a certain domain without having to observe it. Thus where the represented domain might be termed the real-world system, an information system is an artificial representation of that real-world system, as perceived by somebody, built to

54 / 66


enable information processing functions. (Wand, 1996) therefore challenges me to re-engineer Conceprocity starting from a clear ontological stance.

(Wand and Weber, 2002) set out a framework for research on conceptual modelling in connection with information systems which has four main components:

Table 8 Conceptual Modelling Framework Elements (based on (Wand and Weber, 2002, p.364))

Element Meaning Status in Conceprocity 1.01. Conceptual-

modelling grammar

Provides a set of constructs and rules that show how to combine the constructs to model real-while domains.

Largely complete. We need to give further consideration in particular to properties, since the current representation (sub-concepts) consumes too much space on the page. Note that we have yet to define the meta-model (Rosemann and Green, 2002) for Conceprocity.

2. Conceptual-modelling method

Provides procedures by which a grammar can be used. Such a method needs to prescribe how to make observations of a domain into a model of the domain.

The method is documented, but for now only in the form of a PowerPoint presentation.

3. Conceptual-modelling script

A script is the product of the conceptual modelling process.

A Lucidchart template exists and this forms the basis of each script.

4. Context The context is the setting in which conceptual modelling occurs and in which scripts are subsequently used.

The initial context of use has been identified and some scripts have already been produced. Conceprocity does not yet possess easy means to produce and maintain a dictionary of the objects it contains, nor any metrics (but see section 5.2).

Macgilchrist challenges me to a more formal knowledge representation approach both for my models and my meta-models; he suggests first order logic in the form of Horn clauses supported by a Prolog implementation – see http://www.coli.uni-saarland.de/projects/milca/courses/comsem/html/index.html.

I am rejecting this possibility for the time being because first-order logic is inaccessible to most of the research volunteers that I might reasonably expect to be able to work with in the near future. For now, I will take a much more pragmatic stance. Conceprocity is what it is because I sought for and have built a pragmatic knowledge representation scheme based on typed concept maps. It is a means to an end. That end is to complete my Ph.D. in the shortest possible time scales and of course to the highest possible standard commensurate with those timescales. As David Weir has put it in conversation: "focus, focus, focus". As I use Conceprocity over the next few weeks to begin to model my own personal knowledge management system, as I use it to structure my final literature review, as I work with my MSc and MAIB students, as I experiment with Conceprocity in

55 / 66

http://www.coli.uni-saarland.de/projects/milca/courses/comsem/html/index.html

http://www.coli.uni-saarland.de/projects/milca/courses/comsem/html/index.html


the classroom, but above all as I use Conceprocity to make explicit the existing personal work systems of research volunteers and to suggest improved personal work systems to them: I will also gradually be building up requirements for Conceprocity 2.0.

6.4 Learning by enquiry: some parallels with Checkland’s LUMASIt is clear that concept maps lack ontological truth value in so far as they are only ever artificial artefacts that present an observer’s point of view concerning what is.

56 / 66


Figure 24 Illustrative summary of some of our propositions

We would comment that this diagram illustrates an inner learning loop as the researcher engages with perceived reality in accordance with some research methodology. She or he learns in a problem-focussed way as (s)he uses methods in an applied methodology. Just as (Argyris, 2000) describes double loop learning in organisations, we suggest that there potentially exists also an outer loop by means of which the researcher may learn at the more profound level described by

57 / 66

Inner learning loop

Outer learning loop


Peter Checkland. (Checkland, 2000) presents (inter alia) LUMAS, Learning for a User by a Methodology-informed Approach to a problem Situation. Taking as his definition of methodology ‘a body of methods used in a particular activity’, Checkland suggests that a user knowledgeable about a methodology perceives a problem situation and uses the methodology to try to improve it. The methodology as a set of principles is converted by the methodology user into a specific method which the user feels to be appropriate for this particular situation at this moment in its history:

“The user U, appreciating a methodology M as a coherent set of principles, and perceiving a problem situation S, asks himself (or herself): What can I do? He or she then tailors from M a specific approach, A, regarded as appropriate for S, and uses it to improve the situation. This generates learning L, which may both change U and his or her appreciations of the methodology: future versions of all the elements [of] LUMAS may be different as a result of each enactment of the process shown.” (Checkland, 2000)

Figure 25 Checkland's LUMAS model Source: (Checkland, 2000)

Checkland stresses that it is not the methodology which leads to improvement. It is the user as (s)he benefits from using the guidelines, as (s)he takes the formally defined methodology M to create or tailor A, the actual, user- and situation-specific approach adopted to the Real –world problem R that (s)he perceives a concern for.

Asked recently in an interview by Frank Stowell (Stowell, 2013), whether he sees the “systems approach” as “a scientific methodology”; if so, how does it guide

58 / 66


scientific inquiry, in your opinion? If not, how would you describe the relationship between a “systems approach” and scientific inquiry?”, Peter Checkland has replied:

“

As for the phrase ‘a systems approach’ I see it as being the name of any epistemology which encompasses the idea ‘system’; defined as the name of the concept of an adaptive whole which can adapt and survive in a changing environment. It thus has only epistemological, not ontological status. This is crucially important for the Systems Movement, this difference between Natural Science and so called ‘Social Science’. Thus, Marx has a theory of history, and his ideas change history, which is not law governed. On the other hand Copernicus and Galileo have different theories concerning whether our local universe is sun-centred or earth-centred; but these ideas can have no effect whatsoever on what is the case out there in the universe, which is law governed. What this means for a ‘systems approach’ is that if it engages with human and social phenomena it can develop only useful epistemology, not discover laws.

” (Stowell, 2013, p.54)4

Thus we suggest the existence of problem-focussed or situational learning – using methods in an applied methodology; and higher-level learning – which will manifest itself in a deepening appreciation of methodology and a concern to develop it further in action. We also suggest the possibility that the outer loop corresponds more-or-less directly to the inquiring / learning cycle of Checkland’s Soft Systems Methodology SSM.

6.5 An application to student learningAccepting that his approach to concept mapping has as a primary justification Checkland’s Learning for a User by a Methodology-informed Approach to a problem Situation, the author has set a group of postgraduate students on a programme of international communications and digital marketing who are, in a particular module, studying the principles of e-commerce: the task of individually choosing an academic article relating to business information systems or to e-commerce or e-business. The individual assignment is to choose one article from a leading information systems or e-business or e-commerce journal, to read it very carefully, to identify and map the concepts within it, to summarise the article, and then to identify and discuss the practical consequences and usefulness of the theory in the actual practice of e-business. So each student on the module will

4 Checkland might perhaps resist ontological commitment. See http://en.wikipedia.org/wiki/Ontological_commitment.

59 / 66

http://en.wikipedia.org/wiki/Ontological_commitment


create a single summary document or presentation and then upload the summary and implications to cloud-based storage so that all students can see the results of the work of others. When mapping the concepts students must use the Conceprocity concept process mapping approach. The results and the learning will both be evaluated in the near future.

6.6 Complementary approaches to concept mapping as part of a mixed-methods research approachThe author’s current research is at heart a multi-methodology – cf. (Avison et al., 1998) - mixed-methods and inherently exploratory approach to a research question which can be simplified to:

“What is the contribution of personal information management systems PIMS to the working model and personal work system of knowledge workers?”

Mixed methods research mainly refers to quantitative and qualitative research in differing mixes. For an introduction to the issues, see (Ågerfalk 2013). (Goldkuhl 1995) presents a Habermasian view of information and action which is in contrast both to pragmatism as seen in (Ågerfalk 2010) and critical realism as seen in (Mingers et al. 2013) and (Zachariadis et al. 2013).

Table 9 indicates how two contrasting forms of concept mapping are used in complementary experiments already underway or soon to be started. These two forms of concept mapping (which I believe to be complementary) are:

1. Conceprocity concept-process maps. Conceprocity CAPRI or CAPRILOPE models are the result of conscious analysis and specific design by Conceprocity modellers.

2. Leximancer “fuzzy” concept maps.

Concerning Leximancer: (Smith and Humphreys, 2006) report that the Leximancer system is a relatively new method for transforming lexical co-occurrence information from natural language into semantic patterns in an unsupervised manner. It employs two stages of co-occurrence information extraction—semantic and relational—using a different algorithm for each stage. “The Leximancer system performs a style of automatic content analysis. The system goes beyond keyword searching by discovering and extracting thesaurus-based concepts from the text data, with no requirement for a prior dictionary, although one can be used if desired. These concepts are then coded into the text, using the thesaurus as a classifier. The resulting asymmetric concept co-occurrence information is then used to generate a concept map.” (Smith and Humphreys, 2006, p.262).

60 / 66


Thus what I term Leximancer “fuzzy” concept maps emerge from unsupervised (or, better in practice, semi-supervised) semantic mapping of natural language text. Figure 26 shows the result of a semi-supervised Leximancer analysis of the author's PhD journal (circa 130,000 words):

Figure 26 Fuzzy concept map of the author's PhD journal produced using Leximancer

Leximancer automatically recognises only single-word concepts. Most of my research concerns compound concepts; here is the list used when producing Figure26 5:

• information system (merge with IS)

5 To inform Leximancer of the centrality of such compound terms, it is necessary to Edit Compound Concepts: to manually compound selected concepts via Boolean operators to obtain deeper and more meaningful analysis. Clicking ‘Edit’ at the ‘Compound Concepts’ stage enables this. Subsequently, Compound concepts will automatically appear on the map if they are regular word concepts.

61 / 66


• personal information management system (merge with PIMS)

• work system

• personal work system (merge with PWS)

• action research

• knowledge management (merge with KM)

• knowledge representation (merge with KR)

• personal knowledge management (merge with PKM)

• Personal Information Management (merge with PIM)

Table 9 also evidences the significance of concept mapping in this research work.

Table 9 Experiments planned or underway in the current research of Mark Gregory

Experiment Concept mapping approach1. Analyse my own auto-ethnography using Leximancer

emergent or fuzzy concept maps. This involves using Leximancer to enquire into my auto-ethnographic PhD journal (130000 words).

Leximancer. I will learn how to seed Leximancer with compound concepts (e.g. information system, personal information management, personal information management system) and thus to refine and focus the resultant concept map. An early attempt at this analysis appears as Figure 26

2. Building various text corpora and then analysing them Leximancer; seeking the emergence of significant vocabulary as a fuzzy concept map

Recognised writing concerning personal information management

Seeking evidence of a systems approach in the PIM literature; expecting the null hypothesis

Key information systems literature Seeking evidence of a systems approach in the IS literature; expecting the hypothesis but at a low level of significance

Key literature concerning the epistemology and ontology of personal information management and personal knowledge management

Seeking an emergent vocabulary

3. Analyse my own auto-ethnography using Conceprocity; the outcome will be a directed and synthetic concept map

Conceprocity CAPRILOPE; the outcome expected to be an initial definition of a Working Model

4. Observing the usability and usefulness of Conceprocity mapping used by postgraduate students as a means of understanding and elucidating research articles

Conceprocity CAPRI; perhaps some use of CAPRILOPE; the outcomes expected to be (1) a better understanding of the extent to which these two usage profiles are used and useful to students and probably (2) refinements to both usage profiles

5. Encourage recognised PIM researchers to audit their own personal information management approach in the form of a PIM Audit; to discuss the resultant approaches and perhaps to map some of them

Limited use of Conceprocity CAPRILOPE

6. Mentored action research with a small number of research volunteers

Evaluation will make limited use of Conceprocity CAPRILOPE

62 / 66


ReferencesAckoff, R.L., Gupta, S.K., Minas, J.S., 1962. Scientific method: Optimizing applied research decisions.

Wiley New York.Allen, D., 2003. Getting things done: The art of stress-free productivity. Penguin Books.Argyris, C., 2000. Double loop learning in organizations. Class. Manag. Thought 115–125.Ausubel, D.P., 1963. The psychology of meaningful verbal learning.Ausubel, D.P., 2000. The acquisition and retention of knowledge: A cognitive view. Kluwer Academic Pub.Avison, D.E., Wood-Harper, A.T., Vidgen, R.T., Wood, J.R.G., 1998. A further exploration into information

systems development: The evolution of Multiview 2. Inf. Technol. People 11, 124–139.Baskerville, R.L., 2011. Individual information systems as a research arena. Eur. J. Inf. Syst. 20, 251–254.Basque, J., 2013. La modélisation des connaissances en milieu organisationnel.Behar, R., 1997. The vulnerable observer: Anthropology that breaks your heart. Beacon Press (MA).Boisot, M., McKelvey, B., 2010. Integrating modernist and postmodernist perspectives on organizations:

A complexity science bridge. Acad. Manag. Rev. AMR 35, 415–433.Booch, G., Rumbaugh, J., Jacobson, I., 2005. Unified Modeling Language User Guide, The (2nd Edition)

(Addison-Wesley Object Technology Series). Addison-Wesley Professional.Bunge, M., 1977. Treatise on basic philosophy: Volume 3: Ontology 1: The furniture of the world. Reidel

Boston.Bunge, M.A., 1979. Treatise on basic philosophy. Vol. 4, Ontology II: a world of systems. Reidel Boston.Buzan, T., Buzan, B., 1996. The mind map book: how to use radiant thinking to maximize your brain’s

untapped potential. Plume Books.Checkland, P., 1981. Systems thinking, systems practice. Wiley, Chichester.Checkland, P., 2000. Soft Systems Methodology: A Thirty Year Retrospective. Syst Res 17, S11–S58.Checkland, P., Tsouvalis, C., 1997. Reflecting on SSM: The Link Between Root Definitions and Conceptual

Models. Syst Res Behav Sci 14, 153–168.Chen, P.P., 1976. The entity-relationship model—toward a unified view of data. ACM Trans. Database

Syst. TODS 1, 36.Ciborra, C., Jelassi, T., 1994. Strategic information systems: a European perspective.Codd, E.F., 1971. Normalized data base structure: A brief tutorial, in: Proceedings of the 1971 ACM

SIGFIDET (now SIGMOD) Workshop on Data Description, Access and Control. pp. 1–17.Conant, R.C., Ashby, W.R., 1970. Every good regulator of a system must be a model of that system. Int. J.

Syst. Sci. 1, 89–97.Cook, W.R., Hill, W., Canning, P.S., 1990. Inheritance is not subtyping, in: Proceedings of the 17th ACM

SIGPLAN-SIGACT Symposium on Principles of Programming Languages, POPL ’90. ACM, New York, NY, USA, pp. 125–135.

Daconta, M.C., Obrst, L.J., Smith, K.T., 2003. The Semantic Web a guide to the future of XML, Web services, and knowledge management. Wiley Pub., Indianapolis.

Date, C.J., 2003. An Introduction to Database Systems, 8th ed. Addison Wesley.Davies, J., Studer, R., Warren, P., 2006. Semantic web technologies. John Wiley & Sons.Davis, E., 2001. Knowledge Representation, in: Smelser, N.J., Baltes, P.B. (Eds.), International

Encyclopedia of the Social & Behavioral Sciences. Elsevier Amsterdam/New York, NY, pp. 8132–8139.

Davis, R., Shrobe, H., Szolovits, P., 1993. What is a knowledge representation? AI Mag. 14, 17.Delamont, S., 2007. Arguments against auto-ethnography, in: Paper Presented at the British Educational

Research Association Annual Conference, 5-8 September 2007. Presented at the British

63 / 66


Educational Research Association Annual Conference, Institute of Education, University of London, p. 8.

Denzin, N.K., 1992. The many faces of emotionality: Reading persona, in: Ellis, C., Flaherty, M.C. (Eds.), Investigating Subjectivity: Research on Lived Experience. SAGE Publications, Incorporated, pp. 17–30.

Eden, C., 2004. Analyzing cognitive maps to help structure issues or problems. Eur. J. Oper. Res. 159, 673–686.

Friedman, A., Smiraglia, R.P., 2013. Nodes and arcs: concept map, semiotics, and knowledge organization. J. Doc. 69, 27–48.

Friedman, A., Thellefsen, M., 2011. Concept theory and semiotics in knowledge organization. J. Doc. 67, 644–674.

Gregory, M., 2010. Collaboration and end-user information management tools. Presented at the 16th Americas Conference on Information Systems: AMCIS 2010, Lima, Peru.

Gregory, M., Norbis, M., 2008. Towards a Systematic Evaluation of Personal and Small Group Information and Knowledge Management. Presented at the 5th International Conference on Cybernetics and Information Technologies, Systems and Applications: CITSA 2008, Orlando, FA.

Gruber, T.R., 1993. A translation approach to portable ontology specifications. Knowl. Acquis. 5, 199–199.

Guarino, N., 1998. Formal ontology in information systems. IOS Pr.Guarino, N., 2006. Ontology and Terminology: how can formal ontology help concept modeling and

terminology? Presented at the EAFT-NordTerm ws on Terminology, Concept Modeling and Ontology, Vaasa.

Guarino, N., Poli, R., 1995. Formal ontology, conceptual analysis and knowledge representation. Int. J. Hum. Comput. Stud. 43, 625–640.

Hartshorne, C., Weiss, P., Burks, A., 1931. Collected papers of Charles Sanders Pierce (Volumes 1-8). Cambridge, MA: Harvard University Press.

Hevner, A.R., 2010. Design Research in Information Systems: Theory and Practice. Springer.Hevner, A.R., Chatterjee, S., 2010. Design Science Research in Information Systems. Des. Res. Inf. Syst. 9–

22.Hevner, A.R., March, S.T., Park, J., Ram, S., 2004. Design science in information systems research. MIS Q.

28, 75–105.Hjørland, B., 2009. Concept theory. J. Am. Soc. Inf. Sci. Technol. 60, 1519–1536.Hjørland, B., Nicolaisen, J., 2005. The epistemological lifeboat. Epistem. Philos. Sci. Inf. Sci. Available

Www Db Dkjnilifeboathome Htm.Inhelder, B., Piaget, J., 1955. De la logique de l’enfant à la logique de l’adolescent: essai sur la

construction des structures opératoires formelles. Presses universitaires de France.Jackson, K.M., Trochim, W.M.K., 2002. Concept mapping as an alternative approach for the analysis of

open-ended survey responses. Organ. Res. Methods 5, 307–336.Jacob, E.K., 2004. Classification and categorization: A difference that makes a difference. Libr. Trends 52,

515–540.Jones, W.P., 2007. Keeping found things found: The study and practice of personal information

management. Morgan Kaufmann Pub.Kowalski, R., 1974. Predicate logic as a programming language. Inf. Process. 74, 569–574.Macgilchrist, R.S.R., n.d. Learning and the Semantic Web.Minsky, M., 1975. A Framework for Representing Knowledge:, in: Winston, P.H. (Ed.), The Psychology of

Computer Vision. McGraw-Hill: New York, pp. 211–277.

64 / 66


Novak, J.D., Cañas, A.J., 2008. The theory underlying concept maps and how to construct and use them. Fla. Inst. Hum. Mach. Cogn. Pensacola Fl Www Ihmc Ushttpcmap Ihmc UsPublicationsResearchPapersT HeoryCmapsTheoryUnderlyingConceptMaps Htm.

Paquette, G., 2010. Visual Knowledge and Competency Modeling - From Informal Learning Models to Semantic Web Ontologies. IGI Global., Hershey, PA.

Peyton Jones, S., Blackwell, A., Burnett, M., 2003. A user-centred approach to functions in Excel, in: Proceedings of the Eighth ACM SIGPLAN International Conference on Functional Programming. p. 176.

Polya, G., 1988. How to solve it. Princeton University Press Princeton, NJ.Rocha Souza, R., Tudhope, D., Barcellos Almeida, M., 2010. Towards a taxonomy of KOS: Dimensions for

classifying Knowledge Organization Systems. Presented at the 2010 ISKO Conference, Rome.Rodriguez, N., Ryave, A., 2002. Systematic self-observation. Sage Publications, Inc.Rosemann, M., Green, P., 2002. Developing a meta model for the Bunge–Wand–Weber ontological

constructs. Inf. Syst. 27, 75–91.Sauermann, L., 2005. The semantic desktop-a basis for personal knowledge management, in:

Proceedings of the I-KNOW. pp. 294–301.Sauermann, L., Bernardi, A., Dengel, A., 2005. Overview and outlook on the semantic desktop, in:

Proceedings of the 1st Workshop on The Semantic Desktop at the ISWC 2005 Conference. pp. 1–19.

Scheer, A.-W., 2000. ARIS--business process modeling. Springer.Scheer, A.-W., Thomas, O., Adam, O., 2005. Process modeling using event-driven process chains, in:

Dumas, M., Aalst, W.M.P. van der, Hofstede, A.H.M. ter (Eds.), Process-Aware Information Systems: Bridging People and Software Through Process Technology. John Wiley & Sons.

Smith, A.E., Humphreys, M.S., 2006. Evaluation of unsupervised semantic mapping of natural language with Leximancer concept mapping. Behav. Res. Methods 38, 262–279.

Smith, J.M., Smith, D.C.P., 1977. Database abstractions: aggregation and generalization. ACM Trans Database Syst 2, 105–133.

Sowa, J.F., 1984. Conceptual structures: information processing in mind and machine. Read. MA.Sowa, J.F., 1992. Conceptual graph summary. Concept. Struct. Curr. Res. Pr. Ellis Horwood New York

Lond. Tor. 3–66.Sowa, J.F., 2000. Ontology, metadata, and semiotics. Concept. Struct. Log. Linguist. Comput. Issues 55–

81.Sperry, R.W., 1975. Left-brain, right-brain. Saturday Rev. 2, 30–33.Spry, T., 2001. Performing autoethnography: An embodied methodological praxis. Qual. Inq. 7, 706–732.Stowell, F., 2013. Peter Checkland Interview. Int. J. Inf. Technol. Syst. Approach 6, 53–60.Stowell, F., Welch, C., 2012. The Manager’s Guide to Systems Practice: Making Sense of Complex

Problems. John Wiley & Sons.Thimor, J., Fidelman, U., 1995. Concept-maps, logic and the cerebral hemispheres. Kybernetes 24, 35–51.Trochim, W.M.K., 1989. An introduction to concept mapping for planning and evaluation. Eval. Program

Plann. 12, 1–16.Trochim, W.M.K., 1993. The reliability of concept mapping, in: Annual Conference of the American

Evaluation Association, Dallas, Texas.Völkel, M., Haller, H., 2009. Conceptual data structures for personal knowledge management. Online Inf.

Rev. 33, 298–315.Wand, Y., 1996. Ontology as a foundation for meta-modelling and method engineering. Inf. Softw.

Technol. 38, 281–287.

65 / 66


Wand, Y., Weber, R., 2002. Research Commentary: Information Systems and Conceptual Modeling--A Research Agenda. Inf. Syst. Res. 13, 363–376.

Weick, K.E., Sutcliffe, K.M., Obstfeld, D., 2005. Organizing and the process of sensemaking. Organ. Sci. 16, 409–421.

Wilson, B., 1984. Systems: concepts, methodologies, and applications. John Wiley & Sons, Inc., New York, NY, USA.

Zeng, M.L., 2008. Knowledge Organization Systems (KOS). Knowl. Organ. 35, 160–182.

66 / 66