Academic journal article Journal of the Association for Information Systems

Learning the Differences between Ontologies and Conceptual Schemas through Ontology-Driven Information Systems

Academic journal article Journal of the Association for Information Systems

Learning the Differences between Ontologies and Conceptual Schemas through Ontology-Driven Information Systems

Article excerpt

Absract:

In the traditional systems modeling approach, the modeler is required to capture a user's view of some domain in a formal conceptual schema. The designer's conceptualization may or may not match with the user's conceptualization. One of the reasons for these conflicts is the lack of an initial agreement among users and modelers concerning the concepts belonging to the domain. Such an agreement could be facilitated by means of an ontology. If the ontology is previously constructed and formalized so that it can be shared by the modeler and the user in the development process, such conflicts would be less likely to happen. Following up on that, a number of investigators have suggested that those working on information systems should make use of commonly held, formally defined ontologies that would constrain and direct the design, development, and use of information systems - thus avoiding the above mentioned difficulties. Whether ontologies represent a significant advance from the more traditional conceptual schemas has been challenged by some researchers.

We review and summarize some major themes of this complex discussion. While recognizing the commonalities and historical continuities between conceptual schemas and ontologies, we think that there is an important emerging distinction that should not be obscured and should guide future developments. In particular, we propose that the notions of conceptual schemas and ontologies be distinguished so as to play essentially different roles for the developers and users of information systems. We first suggest that ontologies and conceptual schemas belong to two different epistemic levels. They have different objects and are created with different objectives. Our proposal is that ontologies should deal with general assumptions concerning the explanatory invariants of a domain - those that provide a framework enabling understanding and explanation of data across all domains inviting explanation and understanding.

Conceptual schemas, on the other hand, should address the relation between such general explanatory categories and the facts that exemplify them in a particular domain (e.g., the contents of the database). In contrast to ontologies, conceptual schemas would involve specification of the meaning of the explanatory categories for a particular domain as well as the consequent dimensions of possible variation among the relevant data of a given domain. Accordingly, the conceptual schema makes possible both the intelligibility and the measurement of those facts of a particular domain. The proposed distinction between ontologies and conceptual schemas makes possible a natural decomposition of information systems in terms of two necessary but complementary epistemic functions: identification of an invariant background and measurement of the object along dimensions of possible variation. Recognition of the suggested distinction represents, we think, a natural evolution in the field of modeling, and significant principled guidance for developers and users of information systems.

Key Words: ontologies, conceptual schemas, modeling

I. Introduction

In the conventional conceptual modeling activity, ontologies are either bypassed or lack a formal specification [Frank, 1997]. Guarino [1998] considers that every IS has an ontology that often is not explicit but obscurely embedded in parts of the system. Hirschheim et al. [1995] make a similar point for IS, saying that all data modeling "presumes either implicitly or explicitly some form of modeling of data meaning." And beyond that, the designer's conceptualization may or may not match with the user's conceptualization. One of the reasons for these conflicts is the lack of an initial agreement between users and modelers concerning the concepts belonging to the domain. Such an agreement could be facilitated by means of an ontology. If the ontology is previously constructed and formalized so that it can be shared by the modeler and the user in the development process, such conflicts would be less likely to happen. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed

Oops!

An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.