Gary Templeton and Charles Snyder
Department of Management
Auburn University, 36849-5241, U.S.A.
The paper describes the operation of a methodology used for ontologically specifying the key concepts in the field of Knowledge Management. Ontological specification is of particular interest to knowledge managers because associated methods support top management in the processing of tacit knowledge into a more explicit form. The paper is a part of a larger project designed to utilize ontological processes for the building of a 'best practices database' in the KM field. Implications of such a database involve greater speed in which dynamic and unstructured fields such as KM can develop into a more explicit and transferable form of knowledge.
Since the advent of computing technology, researchers in information systems and particularly in artificial intelligence (AI) have made several attempts at the formalization of a process which can be termed ontological specification. An ontology is the explicit specification of conceptual meaning in a topic area such as Knowledge Management (KM) in this project. The ontology includes a complete vocabulary and logical statements about the terminology, how the terms are defined and how they are related (or not related). The process involves the derivation of more specific meaning from more general concepts. This is one method for providing meaning in a knowledge domain. Its applicability in AI is such that methods can be constructed which attempt to automate the human information processing tasks associated with deducing logical structures, commonalities and relationships in conceptual representations found in textual data. The goal of such methods is the reduction in need for humans to be inundated with the belaboring task of extracting meaning from volumes of text. The need for such systems derives from the proliferation of web-based media, which has caused a dramatic increase in organizational access to business intelligence in textual form. The application of such structuring processes could help greatly in the growing area of KM, an organizational function where knowledge is considered to be the strategic focal point for competitive advantage.
A problem that organizations are faced with is that of increasing access to masses of strategically critical contextual and conceptual information. This coincides with limited means of automated support in the form of computer-based information technology for the capturing and organization of important knowledge. This paper provides some insights on the idea of ontological specification (or concept structuring) by depicting the application of one such method in the area of KM. Thus, we describe a tool that can greatly aid in the process of KM in its own field.
Although the field of KM has just recently started to take off, the prospect of gathering all of the information published in the area is formidable. This is especially true given the enormous growth in the amount of information published each month on the topic area. The immaturity, yet dynamic development of the field of KM means that the field has achieved little structure in terms of frameworks, commonly held beliefs, taxonomies, and terminology. This has made the field an excellent target for the application of a method for its structuring.
Previous Work in Structuring Tacit Knowledge
Much work has been done with the goal of providing algorithmic approaches to extracting knowledge from empirical data (whether qualitative or quantitative). Past work in the structuring of numeric data is vast, including the entire field of statistics, graphing methods, and obscure methods, such as the symbolic-oriented FACES concept (Bruckner, 1979). The latter involves an analysis resulting in characterizations about numeric data portrayed in the form of illustrated human "faces".
Attempts at structuring of qualitative text data has a long tradition, beginning with the study of language structures (parts of speech, grammar, etc.), and more recently, proprietary text processing methods made possible from the use of computer-based information technology.
A rudimentary form of IT-enabled proprietary methods might include the counting of each different word appearing in a text. The development of such a database enables the objective development of an argot by using the most commonly used terms found in the analysis. The argot can then be used subjectively to decide which themes (i.e., technological or managerial terms) are emphasized and what themes are important in a body of text. For example, a researcher can determine if a CEO is more concerned with technical or managerial issues by analyzing which type of words he/she emphasizes in the annual statement. In this way, such a method can serve as a utility function used in the preliminary analysis of a body of text.
There are various forms in which AI programs work to transform English statements into computer-based representations for processing. These methods generally take the form of transferring statements into a objects, actions and actor representations for the output of logical inferences (Schank, 1984). While serving its purpose in uncovering inferences, this method does not have practical use in processing a mass of text.
Perhaps the most advanced attempt at using language structures for management, other than the implications of ontological specification, has evolved in the form of Zachman's (1987) ISA Framework. Zachman used three (later developed into six) possible questions asked about a business system with each segregated into six roles of system analysts. The original framework considered every system to be composed of data ("what"), process ("how"), and network ("where"). Sowa and Zachman (1992) later extended the three question categories to include "who, when, and why" components of systems, meaning that all questions available in the English language are reflected in the model. Implications were that information related to all possible questions about the real-world business system could be represented in the computer.
A convenient idea in information systems literature concerns the applicability of normalized database relations in relating to themes of meaning. The subject nature in which normalized database relations are evaluated relates strongly to judgements concerning the appropriateness of ontology structure. In fact, the two processes are very similar. Both are attempts at capturing meaning using standardized representations of meaning (concepts and subconcepts for ontological specification and relations and elements for normalization). It can be said that both processes are attempts at organizing knowledge in a more straight-forward manner for an end user (or agent) of the knowledge acquired.
The Ontological Concept-Specification Process
Definition of an Ontology
An ontology represents a specifying scheme of concepts which holistically describes some topic. For our purposes, it can yield declarative knowledge about the structure and processes related to the KM concept. It can serve as a formal vocabulary for researchers, instructors, students, and practitioners in the KM community of practice and includes logical descriptions of the items, relationships between items, and how items cannot be related (Gruber, "What is an ontology?").
The utility of specific terminology is a greater stimulation in more refined concepts in the field, which can result in further formalization of the topic area and structuring of decision-making processes in the field. Further, the KM ontology can be used by researchers to uncover what subconcepts are and are not related when such relationships are of interest. This process of greater formalization inevitably results in prescriptive associations in academic fields, which means that methods can be prescribed for various contingencies found in practice. Ontologies can be viewed as a knowledge-based communications technology in that a greater ability to represent and communicate knowledge about a concept is possible as more terms and relationships are uncovered in the ontological specification process. With this definition of ontological specification at hand, we can see that the process itself can serve as a subconcept in the KM ontology (see table), a scenario which adds to the implications of this research.
The Utility of an Ontology
In viewing the ontological specification process as a knowledge-based technology, researchers at Stanford University are at the forefront. There, a web-based ontology building application resides, allowing users to build ontologies of any subject matter. The system, called Ontolingua, serves as a knowledge acquisition laboratory for the AI faculty and supports the standardization of knowledge structures transferable to intelligent software modules. The system has been used to build ontologies mainly in the field of AI, but some in areas which can be directly linked to KM: Bibliographic-Data, Documents, Job-Assignment-Task, User, Design, and Domain.
The many-to-many relationship in the ability of the application to receive ontology-building requests and to be able to transfer to varying software environments means that the ontological specification process has received ontological attention. The ontology engineering perspective has yielded a meta-ontology which can be used across any knowledge domain, much of which can be found in the work of Gruber (1992).
The Ontological Specification Process
Levels of specification during the ontological specification process are delineated as shown in Figure 1. While not all ontological specification methodologies are the same, all can be said to share common steps. These steps are 1) selection of the topic area, 2) delineation of concepts, 3) transfer to a usable medium, and 4) use of concepts. Operationalization of the steps depends on the purpose of the ontology. For example, the purpose of the Stanford ontology procedure involves capturing the knowledge for transfer into rigid data structures, which requires further analysis of the definitions of knowledge elements. The current study uses the ontological process as a way of marking literature for future literature reviews of the KM field. The current methodology was structured to accommodate this purpose, in the form of 1) selection of the KM topic, 2) delineation of concepts (as illustrated in Figure 1) by literature review, 3) denotation of concepts in the literature, 4) transfer to database format, and 5) use of database in research.
The Ontological Specification of KM Illustrated
Selection of the KM Topic
Selection of the topic of KM for ontological analysis was done due to its place on the research continuum. A review of the literature showed that an overwhelming amount of the knowledge available about KM was descriptive in nature. This meant that most effort had been justifyingly aimed toward the definition and uncovering of key concepts in the field. Very little academic empirical work had been conducted except for conceptual "blue sky" (normative) works using examples of KM utilization in the field (as in Sanchez and Mahoney, 1996). Thus, the field was a prime target for formal structuring methods such as ontological and taxonomic specification.
Delineation of Concepts by Literature Review
Operation of ontological specification should be seen as an iterative and subjective behavior of the agent operator and heavily dependent on operator learning. The classic iterative control process is used throughout ontology development, following the steps of 1) setting an ideal, 2) setting standards, 3) evaluation of feedback data, 4) changing operations or the ideal in perpetual cycles (Newman, 1975). Deciding on an exact ontology cannot be accomplished in such a new and volatile field as KM, where expected discoveries in the field will relate to a redefinition of the ontology. Defining and placing relevant subconcepts in an ontology is done based on some purpose, such as 1) to define and refine a researcher's interests, 2) to capture common themes in a body of literature and 3) to organize experiential knowledge about a research topic.
Thus, the delineation of the KM concept does not mean that KM will be the most significant or important field in the analysis. It means that as we decompose the concept, other related concepts should be included in the study of KM (see Figure 1). For example, executive information systems, telecommunications and database are more mature and may be more important in the eyes of managers, but they may relate in a strong way to KM. The relationship between these important concepts and KM is of pertinent concern to users of the ontology and thus justifies its existence in the scheme.
Figure 1: Delineation of Levels in the Ontological Specification Process
Level of Description
Concept (KM) Specification
Concept (KM) Specification
Concept (KM) Specification
Concept (KM) Specification
Concept (KM) Specification
|Example from Current
III. KM Operation
D. How KM impacts the organization (OI)
1. Organizational change and KM ()
a. How technology can support KM (TECH)
i. Learning Systems (LS)
ii. Best Practices databases (BEST)
iii. Organizational Memory Info. Sys. (OMIS)
iv. Networking and KM (NET)
An initial review of the KM literature revealed three categories of knowledge in the literature associated with the KM concept. These categories are Resource Meta-Data, KM Description and KM Operation (see Table 1).
Meta-Knowledge involves the entomological view, which includes knowledge about sources, states, structures, processes, histories, and evaluations of knowledge about KM, including the current paper. In most fields, Meta-Knowledge is important for researchers but of little value to practitioners. However, the understanding of Meta-Knowledge is the goal of one who studies KM. This category of knowledge is where the objectives of practice and academia coincide.
The KM Description concept involves descriptive knowledge about the field. This type of knowledge is emphasized early in the development of academic fields and is concerned with the defining of key issues, terms, and the history of the field. Writers of KM literature have emphasized KM Description knowledge for better understanding of the meanings of the concept and related terminology.
KM Operation involves the structure and processes associated with the topic. We use this type of knowledge to depict for management the way KM should be aligned and operate. This type of knowledge has been relatively rare in KM literature because the field has concentrated on describing the field (as in KM Description). Work in the area of operationalizing KM will lead to prescriptive knowledge about what causes, effects and contexts are important in operating the organizational KM effort. Thus, this type of knowledge is the goal of academic research and study about the field and is the direction KM is currently heading.
It is important to understand the three categories of KM knowledge posed in the framework. The nature of Meta-Knowledge is descriptive, as in the KM Description category. However, the former is used in describing KM knowledge while the latter is used to describe KM. The KM Operation category is where we learn to competently manage knowledge and its processes.
Table 1 shows how KM can be delineated into more specifying subconcepts. For example, the early development stage of the field results in the need for researchers and practitioners to have agreed-upon definitions and a foundation for terminology. Hence, the consideration of the Definition of KM idea has become important and was placed as a direct subconcept of the KM Description concept in the ontology. The concept was derived by the consistent efforts by several 'guru's' offering varying definitions in the KM literature.
Table 1: The KM Ontology
|I. Resource Meta-Data
A. Source type (SO)*
B. Study Type (ST)
C. Academic base (AB)
D. Empirical Support (EM)
II. KM Description
F. History of KM (HIS)
A. Definition of KM (DEF)
B. KM characteristics (CH)
C. How to determine the presence of KM (PRES)
D. Examples of KM and its Absence (EX)
E. KM Architecture (ARCH)
III. KM Operation
A. Processes of KM (ACT)
1. Determining info requirements during KM (IR)
2. Knowledge Acquisition (KA)
3. Data Management in KM (DM)
4. Processing/Transforming Knowledge (PROC)
5. KM and GST (GST)
6. Organizational Learning (OL)
7. Organizational Memory (OM)
B. Why KM is needed (NEED)
Control Theory and KM (CT)
C. Knowledge as Intangible Asset (IA)
1. Knowledge capital theories (KT)
2. Knowledge creation (KC)
3. Intellectual Capital Management (ICM)
D. How KM impacts the organization (OI)
1. Organizational change as related to KM ()
a. How technology can support KM (TECH)
i. Learning Systems (LS)
ii. Best Practices Databases (BEST)
iii. Organizational Memory Info. Sys. (OMIS)
iv. Networking and KM (NET)
b. KM culture (CULT)
2. Organizational performance (OP)
E. Organizational use of KM (OUSE)
F. Benefits of KM (BEN)
G. Factors effecting quality KM effort (FEQ)
1. Implementation of KM (IM)
2. Evaluation of KM (EV)
3. Characteristics of the Knowledge Manager or Group (CHAR)
* parentheses indicate subconcept tag used in text denotation process
The ontology in Table 1 also shows the decomposition of the KM Operation subconcept of KM. KM Operation literature was found to be described by seven subconcepts relating to the issues associated with KM practice. One intriguing facet of KM Operation is the Processes of KM subconcept, relating to the activities associated with KM practice. These activities may be classified as operational or managerial activities, but were explicitly mentioned in the literature as one of the seven categorizations shown. Further decomposition leads to the Processing/Transforming Knowledge subconcept, which includes practices such as Ontological Specification and the textual processing methods mentioned previously. This ontology shows the congruence in purpose between academics and practitioners in the KM field.
The previous description of the method of concept delineation highlights several problematic issues in the derivation and selection of subconcepts. First, the description points out that a methodology operator must subjectively select from competing ideas the most appropriate and pertinent subconcepts to be placed in the ontological structure. An important principle is that the operator must have knowledge in the area of interest due to the potential for researcher bias in selection. In his engineering approach to ontological specification, Gruber (1993b) refers to this problem as "encoding bias" and uses it as a measure of ontology quality.
Denotation of Concepts in the Literature
For immediate communication to a literature reviewer about what concept meaning is associated with a given set of text, the specification operator simply denotes physically in the literature using concept-associated tags such as those shown in parentheses in Table 1. Notation tag creation is done with two purposes in mind: 1) to support the learning curve parameters of the denoting specialist and 2) to communicate to a notation user about meanings in the text. Denoting text can become extremely tedious without the utilization of communicative tags. For instance, the current methodology was initially employed using numeral tags until a more descriptive variable name approach evolved.
Denotation criteria, the standard by which each reviewed text set is evaluated, is an important consideration and should be documented for all denoting specialists employed on a specification project. Well communicated and implemented evaluation criteria can result in less risk of subjective bias and other problems in the ontology. For example, the researcher denoting the current project on KM would document explicit knowledge ("the statements must contain two of the three words in related statement or set of statements") about the criteria by which Knowledge Capital Theories are tagged with the KCT.
The relationship between denotation tags and textual data items is many-to-many. This means one instance of text can have many tags and one tag can have many instances of text in a body of dialogue. Figure 2 shows specific examples of each, and depicts the complexities which can arise in extracting meaning from textually represented knowledge.
Figure 2: Examples of Denotation in KM Literature
(text extracted from: Amidon and Skyrme, 1996)
|The report's main conclusion is that effective management of knowledge
will be a core competence NEED
that most organizations will need to develop to succeed in tomorrow's dynamic global economy.
Many examples were found of companies who had achieved business growth, reduced costs, faster
time-to-market, innovative products and services, through the systematic application of knowledge GST
management processes. Some examples:
- Dow Chemical have turned unexploited knowledge in its patents into cost savings and a growing BEN
licensing revenue stream.
- Thomas Miller, a London-based manager of mutual insurance companies, share their global
expertise through online systems and learning centers to improve customer service. TECH
- Steelcase, an office equipment company, through applying knowledge or the workplace have
created new product lines that have transformed the nature of the company to one of providing
- BP have used videoconferencing effectively to reduce oil well stoppages, by bring(ing) knowledge EX
to bear quickly on problems.
- Price Waterhouse have created a global knowledge base, that enhances their ability to bring world-
class best practice to deliver customer solutions.
Transfer to (Relational) Database Format
Use of denoted data, stored as shown in Figure 2, can be tedious and time-consuming in knowledge work like literature reviews. For this reason, and due to advances in relational database technologies, it can be said that conversion to a relational format is desired (and is coincidentally a natural process). Figure 3 shows the result of this conversion, a relation logically depicting which literature work is associated with a given concept.
Figure 3: Relational Representation of Denotation Instances
Conversion can be done with or without further analysis, which presents the opportunity for a KM specialist to convert tagging instances to either of the four data types (nominal, ordinal, interval, or ratio). For example, Figure 3 shows that the specialist may input ratio data which may represent the number of times History (HIS) is tagged in a specific work. It is shown that Definition of KM either does (DEF = 1) or does not (DEF = 0) exist in a given title and is therefore represented nominally. Qualitative data in this case is more appropriate for Definition of KM since multiple definitions of KM usually do not usually appear in text sources. The definition of the data type of a concept can be important, because qualitatively defined concepts have implications for the range of capabilities associated with hypertext links between concepts and text instances.
More elaborate examples of concepts where a qualitative design is appropriate are those of Study Type (ST) and Empirical Support (EM). Study Type may be coded as either a field, case, or "blue sky" study upon conversion from denotation for example. These examples show the richness by which denotation can be captured in relational format but also the inevitable reliance the intuitive feel of knowledge workers operating the methodology depicted.
Use of Database in Research.
The storing of denotation in relational form implies that normalization decomposition procedures can be performed on the data in the production of a KM production database. The goal of such a database is to relate a knowledge manager to literature corresponding to good and bad practice in the field, a KM Best Practices Database (KM-BPDB). A data-driven approach to using the KM-BPDB would involve sorting values to find where knowledge about a particular concept exists in the literature. A researcher on KM would use this query to build one ontological specification of the field or to discover associations and terms much faster than with traditional research approaches. For example, the user may wish to review all Definitions of KM (shown in Table 2) for further ontological refining.
Table 2: Some Definitions of KM
(source of definitions: The Knowledge Management Forum, 1997)
|KM 'Guru' Source
||Definition of KM
||the management of the organization towards the
continuous renewal of the organizational knowledge base - this means e.g.
creation of supportive organizational structures, facilitation of organizational
members, putting IT-instruments with emphasis on teamwork and diffusion
of knowledge (as e.g. groupware) into place.
||an audit of "intellectual assets"
that highlights unique sources, critical functions and potential bottlenecks
which hinder knowledge flows to the point of use. It protects intellectual
assets from decay, seeks opportunities to enhance decisions, services and
products through adding intelligence, increasing value and providing flexibility.
||Knowledge Management is the collection of processes
that govern the creation, dissemination, and utilization of knowledge
||the art of creating value from an organization's
||focusing on determining, organizing, directing,
facilitating, and monitoring knowledge-related practices and activities
required to achieve the desired business strategies and objectives.
A goal-driven approach may include powerful drill-down capabilities built into the database in the form of querying languages, embedded scripts and external applications. For example, we may wish to review all articles containing both KM Characteristics (CH) and How Technology Can Support KM (TECH) using SQL capabilities. Management can use the ontological schema to segment work activities in a knowledge model, so that practices and technologies can be researched for a given work setting (much like how software support personnel work currently). In the face of enormous growth in online textual data, these examples show how the process of transforming ontologically specified elements to relational format can support the conceptual decision-making activities of strategic management. Last, systems can be built with capabilities to perform ontological specification, denotation, and transfer to standard relational data structures, and processing of electronically held textual knowledge.
Although a majority of the KM ontology depicted is concerned with prescriptive concepts in the KM field, a majority of the available literature remains descriptive (estimated at about approximately 90%). Appropriately, most literature on KM at this point is available from non-academic sources, as we know very little about variables and constructs in the field. Where no cause-effect relationships have been found, most literature is of the "blue-sky" nature where we assume the normative case and the field's position on the research continuum is justifiable in this early stage of development.
The point of this study was to show that ontological specification can aid researchers in providing structure in its own field, Knowledge Management. The prospect of transferring ontologically derived concepts to databases is very appealing and feasible. It has long been held that databases are much easier and efficiently created, operated and maintained when the data is held in logical structures. The process of normalizing a relation has this purpose and is relevant when the data is held in relational format. Thus, research is needed in the area of proper normalization of such an ontologically derived relation. It is the contention of this research that normalization will follow a very similar decomposition process as that found in the ontology presented and new elements of the field can be uncovered in that process. Further, the notion that an ontological notation scheme should consist of unique values suggests that a given ontology lends itself to the application of a production database scheme, where concepts are records and properties can be generated that describe important dimensions of the field.
Knowledge about KM and ontological process can speed the specification - automation process which has enormous implications for supporting work associated with conceptual knowledge (in unstructured domains of problem solving). The process of considering and culling ideas for positioning on a KM ontology is in itself an important area of research in Knowledge Management. This leads to the postulate that ontological specification is a very intricate part of KM and that its study is vital in understanding KM and organizational learning.
Amidon, Debra M. and Skyme, D.J. 1996. Business Intelligence (newsletter). Published by Business Intelligence, Ltd., London.
Bruckner, Lawrence A. 1979. The interactive use of computer drawn faces to study multidimensional data, Department of Energy, Los Alamos, N.M.
Gruber, T.R. Accessed 5/12/97. What is an ontology? Available from the WWW at: http://www-ksl.stanford.edu/kst/what-is-an-ontology.html.
Gruber, T.R. 1992. Toward principles for the design of ontologies used for knowledge sharing. In Nicola Guarino, Ed., International Workshop on Formal Ontology, Padova, Italy.
Gruber, T.R. 1993a. A Translation Approach to Portable Ontology Specifications. Knowledge Acquisition, 5(2):199-220.
Gruber, T. R. 1993b. Toward principles for the design of ontologies used for knowledge sharing. Originally in N. Guarino & R. Poli, (Eds.), International Workshop on Formal Ontology, Padova, Italy. Revised August 1993. Published in International Journal of Human-Computer Studies, special issue on Formal Ontology in Conceptual Analysis and Knowledge Representation
Gruber, T. R., & Olsen, G. R. 1994. An ontology for engineering mathematics. In J. Doyle, P. Torasso, & E. Sandewall (Eds.), Fourth International Conference on Principles of Knowledge Representation and Reasoning, Gustav Stresemann Institut, Bonn, Germany, Morgan Kaufmann.
Newman, William H. 1975. Constructive Control; Design and Use of Control Systems, Prentice Hall, Englewood Cliffs, N.J., pp. 174.
Schank, Roger. 1984. The Cognitive Computer, Addison-Wesley, Reading, Massachusetts, pp. 268.
Sowa, J.F. and J.A. Zachman. 1992. Extending and Formalizing the Framework for Information Systems Architecture. IBM Systems Journal 31 (3): 590-616.
Sanchez and Mahoney. 1996. Modularity, Flexibility, and Knowledge Management in Product and Organization Design. Strategic Management Journal 17 (Winter Special Issue): 63-76.
The Knowledge Management Forum. Accessed 5/11/97. What is Knowledge Management? Available on the WWW at: http://revolution.3-cities.com/~bonewman/what_is.htm.
Zachman, J.A. 1987. A Framework for Information Systems Architecture. IBM Systems Journal 26 (March): 276-92.