Conceptual modeling for data and knowledge management

https://doi.org/10.1016/S0169-023X(99)00049-XGet rights and content

Abstract

In order to exploit knowledge embedded in databases and to migrate from data to knowledge management environments, conceptual modeling languages must offer more expressiveness than traditional modeling languages. This paper proposes the conceptual graph formalism as such a modeling language. It shows through an example and a comparison with Telos, a semantically rich knowledge modeling language, that it is suited for that purpose. The conceptual graph formalism offers simplicity of use through its graphical components and small set of constructs and operators. It allows easy migration from database to knowledge base environments. Thus, this paper advocates its use.

Introduction

Information technologies are used to help organizations manage their business processes. More specifically, databases were designed to support the data storage and retrieval activities related to information management. Because database management systems provide efficient task support to data intensive activities, tremendous gain in productivity was accomplished using these technologies. Furthermore, these data often take part in higher cognitive activities like strategic planning, process validation and decision making, which are key activities (profitability wise) not only in the day-to-day operations of an organization but also in its medium and long-term development. Consequently, the next generation of task support systems aims at providing direct support to these knowledge-dependent activities.

This knowledge is generally related to different aspects of the management of an organization: its processes, customers, suppliers, markets, activities, past expertise, policies, guidelines, constraints, and so on. The storage, retrieval and processing of this type of information go way beyond available database technology; it calls for knowledge management and information systems. The blend of the different types of knowledge required by the management activities of an organization is often called corporate knowledge. When it is confined to some knowledge management system, it is then called a corporate memory.

In order to model this type of information systems, traditional conceptual modeling languages like the entity-relationship (ER) formalism or UML are very limited due to their lack of semantic expressiveness. Therefore, knowledge modeling languages like NIAM [21] or Telos [20] must be considered. In this paper, we propose a well-established knowledge representation formalism, the conceptual graph (CG) formalism [23], as the representation language used in the modeling activities pertaining to the development of information systems. Through an example and a comparison with Telos, a well-founded and very expressive knowledge modeling language [22], we show that the CG formalism has very attractive features that make it a prime candidate for such task. Through its simplicity and graphical constructs we see the CG formalism as being a simple-to-use knowledge modeling language. Through its expressiveness we show that it produces a complete model of the application domain. Through its formal semantics, it offers deduction capabilities that provide executable specifications helpful to the knowledge engineer. Through its inference capabilities, it offers sophisticated query mechanisms that will provide high-level task support not only to the knowledge engineer but also to the end-user.

As pointed out by Mylopoulos et al. [20], when it comes to the development of information systems: “the primary responsibility of any language intended for the task is to be able to formally represent the relevant knowledge”. Deduction capabilities are needed to validate the model; completeness of representation is important to its validity. Therefore, we totally support this point of view. Furthermore, we add that the secondary responsibility of any conceptual modeling language is to be easy to use and to interpret. Though technological decisions are made at some relatively high level in an organization, technological integration may not be so easy to achieve: individuals at the other end of the scale often validate and implement these technological choices. Also, in terms of profitability, technology integration must be as painless as possible. Simple yet effective technology must be sought. In this paper, through a comparison with Telos, we show that the CG formalism meets these two fundamental responsibilities and is therefore a prime candidate for the conceptual modeling activities pertaining to the development of a knowledge management and information system. Section 2 gives a brief overview of basic conceptual modeling requirements and introduces a simple example using Telos. In Section 3 we present the CG version of the example of Section 2 and proceed to a comparison with Telos. Section 4 discusses related work. Section 5 summarizes our findings. Section 6 concludes.

Section snippets

Conceptual modeling for knowledge-oriented applications

The choice of a modeling language has tremendous impact on the completeness of a conceptual model. Unfortunately, modeling languages are often caught in the completeness/complexity trade-off. The more expressive they are, the more complexity they suffer from. Complexity is an important factor to be considered when an organization plans its technological development. It adds additional costs in terms of getting acquainted with new technology and subsequently mastering it. Development and

The CG version of the example

The CG formalism1 allows both structural and behavioral aspects of the definition of an object to be represented. Section 3.1 shows how structural aspects can be defined, introducing concepts, referents, relations and functions; while Section 3.2 deals with behavioral aspects,

Related work

The potential of the conceptual graph theory for conceptual modeling and constraint representation has been handled in a few studies [2], [4], [5], [6], [24]. In [4], the author argues that conceptual graphs offer modeling features that make them useful as a canonical data model, and hence let them serve to schema description, integration, and transformation from a source model (e.g., entity-relationship model) to a target one (e.g., NIAM). We totally agree with that statement.

In [24], dataflow

Discussion

The aim of this paper was to present the CG formalism as a prime candidate for conceptual modeling tasks, especially for knowledge management and information systems. To reach that goal, a comparison was sketched throughout the paper between the CG formalism and Telos, a semantically rich knowledge modeling language. In Section 2, certain requirements were laid out: completeness of representation (for structure, behavior, time, constraints and contexts), foundation on formal semantics (for

Conclusion

The objective of this paper was to outline the potential that the CG formalism has to offer with regard to conceptual modeling, especially since knowledge management is becoming more and more a real preoccupation in industry nowadays. We feel that the next generation of database systems will soon include knowledge intensive applications, and that the migration of conceptual modeling tasks from traditional conceptual modeling languages to knowledge based technology will eventually become an

Acknowledgements

The authors wish to thank the R&D Division of the DMR Consulting Group for implementing most of the ideas presented in this paper, thus choosing the CG formalism as the representation technology behind their process management tool. This tool was developed to provide task support to their consultants who need to use these processes on a daily basis. This testbed was very important to us to make sure that the ideas conveyed in this paper could be implemented in a real knowledge management

Guy Mineau is an associate professor at the Department of Computer Science of the University Laval in Quebec City since 1990. He received his Ph.D. from the Université de Montréal in 1991. His two main research interests are: the conceptual graph theory and conceptual clustering techniques for complex objects, with direct applications in knowledge modeling, knowledge management, information retrieval and data mining. He is the author or co-author of more than 50 refereed papers. He is a member

References (26)

  • J.F. Allen

    Maintaining knowledge about temporal intervals

    CACM

    (1983)
  • L. Campbell, P. Creasy, A conceptual approach to information systems design, in: Proceedings of the Seventh Annual...
  • P. Creasy, ENIAM: A more complete conceptual schema language, in: Proceedings of the 15th International Conference on...
  • P. Creasy, Conceptual graphs as canonical data model, in: Supplimentary Proceedings of the Second International...
  • P. Creasy, B. Moulin, Approaches to data conceptual modeling, in: Proceedings of the Sixth Annual Workshop on...
  • P. Creasy, B. Moulin, Adding semantics to semantic data models, in: T.E. Nagle, J.A. Nagle, L.L. Gerholz, P.W. Eklund...
  • G. Di Battista et al.

    Deductive entity relationship modeling

    IEEE Transactions on Knowledge and Data Engineering

    (1993)
  • R. Elmasri, S.B. Navathe, Fundamentals of Database Systems, Benjamin/Cummings, Redwood City,...
  • J.W. Esch, Visualizing temporal intervals represented as conceptual graphs, in: Proceedings of the Sixth Annual...
  • D. Lukose, Executable conceptual structures, in: G. Mineau, B. Moulin, J.F. Sowa (Eds.), Lecture Notes in AI, vol. 699,...
  • D. Lukose, G.W. Mineau, A comparative study of dynamic conceptual graphs, in: Proceedings of the 11th Banff Knowledge...
  • G.W. Mineau, Views, mappings and functions: essential definitions to the conceptual graph theory, in: W.M. Tepfenhart,...
  • G.W. Mineau, The representation of dynamic processes using conceptual graphs, Internal Report no. 960611B, DMR...
  • Cited by (41)

    • Big data technologies and Management: What conceptual modeling can do

      2017, Data and Knowledge Engineering
      Citation Excerpt :

      Graph databases are also employed for master data management and version management systems, and complex relationship management in drugs. Conceptual modeling has, since its beginning, focused on the organization of data [4,8,24]. It requires the construction of a conceptual representation of the application domain of an information system [41,43].

    • Extract conceptual graphs from plain texts in patent claims

      2012, Engineering Applications of Artificial Intelligence
      Citation Excerpt :

      Conceptual graph mainly represents structural information. Mineau et al. (2000) have argued that the conceptual modeling languages offer more expressiveness than traditional modeling languages. The conceptual graph can also be converted to a corresponding first-order-logic formula (Amati and Ounis, 2000) in a way that the semantic inferences can be carried out further.

    • Engineering design of green hybrid energy production and supply chains

      2009, Environmental Modelling and Software
      Citation Excerpt :

      The port classification is useful for controlling and managing material and/or energy flow among the different SUs. There are many engineering solutions that are based on robust data and knowledge modeling (Yang et al., 2003; Xindong and Cai, 2000; Mineau et al., 2000; Mili et al., 2001). Energy production chain domain knowledge is constructed based on the proposed energy production chain process modeling (i.e. using POOM).

    • Formal conceptualisation as a basis for a more procedural knowledge management

      2008, Decision Support Systems
      Citation Excerpt :

      The proximity to implementation concepts is also present in knowledge engineering. In fact, some authors consider the conceptual graph formalism to avoid this problem [29]. They explicitly accept, however, that this formalism has important disadvantages for KM such as, for example, its rigidity (i.e., it is based on mathematical semantics) and the lack of a global view of the modelled domain.

    • Open Distance Learning: Fundamentals, Developments, and Modelling

      2022, Open Distance Learning: Fundamentals, Developments, and Modelling
    View all citing articles on Scopus

    Guy Mineau is an associate professor at the Department of Computer Science of the University Laval in Quebec City since 1990. He received his Ph.D. from the Université de Montréal in 1991. His two main research interests are: the conceptual graph theory and conceptual clustering techniques for complex objects, with direct applications in knowledge modeling, knowledge management, information retrieval and data mining. He is the author or co-author of more than 50 refereed papers. He is a member of AAAI.

    Rokia Missaoui received her Ph.D. degree in Computer Science from Université de Montréal in 1988. She joined the department of Computer Science, Université du Quebec à Montréal in 1987, where she is currently a Professor. Her research interests concern object-oriented database modeling and management, data mining and knowledge engineering. She is a member of the ACM and IEEE.

    Robert Godin is a professor at the Computer Science Department of the Université du Québec à Montréal (UQAM) since 1983. He received his Ph.D. from the Université de Montréal in 1986. His current research interests are in the areas of data mining, information retrieval, object oriented software modeling and design, and formal concept analysis. He is the author or co-author of approximately 40 papers in refereed conferences and journals. He participated in many large R&D projects. He was a member of the scientific committee of the Centre de Recherche en Informatique de Montréal (CRIM) from 1988 to 1993.

    View full text