Description Logic and Faceted Knowledge Representation

Uta Priss. Description Logic and Faceted Knowledge Representation. In Patrick Lambrix, Alexander Borgida, Maurizio Lenzerini, Ralf Möller, Peter F. Patel-Schneider, editors, Proceedings of the 1999 International Workshop on Description Logics (DL 99), Linköping, Sweden, July 30 - August 1, 1999. Volume 22 of CEUR Workshop Proceedings, CEUR-WS.org, 1999. [doi]

Abstract

The term ”facet” was introduced into the field of library classification systems by Ranganathan in the 1930’s [Ranganathan, 1962]. A facet is a viewpoint or aspect. In contrast to traditional classification systems, faceted systems are modular in that a domain is analyzed in terms of baseline facets which are then synthesized. In this paper, the term ”facet” is used in a broader mean- ing. Facets can describe different aspects on the same level of abstraction or the same aspect on different lev- els of abstraction. The notion of facets is related to database views, multicontexts and conceptual scaling in formal concept analysis [Ganter and Wille, 1999], poly- morphism in ob ject-oriented design, aspect-oriented pro- gramming, views and contexts in description logic and semantic networks.

This paper presents a definition of facets in terms of faceted knowledge representation that incorporates the traditional narrower notion of facets and potentially fa- cilitates translation between different knowledge repre- sentation formalisms. A goal of this approach is a mod- ular, machine-aided knowledge base design mechanism. A possible application is faceted thesaurus construction for information retrieval and data mining. Reasoning complexity depends on the size of the modules (facets). A more general analysis of complexity will be left for future research.