2009 IdealDownwardRefinementintheELD

• (Lehmann & Haase, 2009) ⇒ Jens Lehmann, and Christoph Haase. (2009). “Ideal Downward Refinement in the EL Description Logic.” In: Proceedings of the 19th International Conference on Inductive logic programming. ISBN:3-642-13839-X, 978-3-642-13839-3

Subject Headings: Refinement Operator, Description Logical Refinement Operator, EL Refinement Operator.

Cited By

• Google Scholar: ~ 58 Citations
• ACM-DL: ~ 13 Citations
• Semantic Scholar: ~ 40 Citations

Quotes

Abstract

With the proliferation of the Semantic Web, there has been a rapidly rising interest in description logics, which form the logical foundation of the W3C standard ontology language OWL. While the number of OWL knowledge bases grows, there is an increasing demand for tools assisting knowledge engineers in building up and maintaining their structure. For this purpose, concept learning algorithms based on refinement operators have been investigated. In this paper, we provide an ideal refinement operator for the description logic EL and show that it is computationally feasible on large knowledge bases.

1 Introduction

The Semantic Web is steadily growing and contains knowledge from diverse areas such as science, music, people, books, reviews, places, politics, products, software, social networks, as well as upper and general ontologies. The underlying technologies, sometimes called Semantic Technologies, are currently starting to create substantial industrial impact in application scenarios on and off the web, including knowledge management, expert systems, web services, e—commerce, e—collaboration,, etc. Since 2004, the Web Ontology Language OWL, Which is based on description logics (DLs), has been the W3C—recommended standard for Semantic Web knowledge representation and is a key to the growth of the Semantic Web.

However, recent progress in the field faces a lack of well—structured ontologies with large amounts of instance data due to the fact that engineering such ontologies constitutes a considerable investment of resources. Nowadays, knowledge bases often provide large amounts of instance data without sophisticated schemata. Methods for automated schema acquisition and maintenance are therefore being sought (see e.g. [5]). In particular, concept learning methods have attracted interest, see e.g. [2,63,11,13].

Many concept learning methods borrow ideas from Inductive Logic Programming including the use of refinement operators. Properties like ideality, completeness, finiteness, piopeiness, minimality and non—redundancy are used as theoretical criteria for the suitability of such operators. It has been shown in [12] that no ideal refinement operator for DLs such as ALC, SHOIN, and SROIQ can exist (the two latter DLs are underlying OWL and OWL 2, respectively). In this article, an important gap in the the analysis of refinement operator properties is closed by showing that ideal refinement operators for the DL EL do exist, which in turn can lead to an advance in DL concept learning.

EL is a light—weight DL, but despite its limited expressive power it has proven to be of practical use in many r[eal—world]] large—scale applications. For example, the Systematized Nomenclature of Medicine Clinical Terms (SNOMED CT) [4] and the GENE ONTOLOGY [18 ] are based on EL. Since standard reasoning in EL is polynomial, it is suitable for large ontologies. It should furthermore be mentioned that EL++, an extension of EL will become one of three profiles in the upcoming standard ontology language OWL 2.

Overall, we make the following contributions in this paper: We

• close a gap in the research of properties of refinement operators in DLs,
• provide an ideal and practically useful refinement operator for EL and
• show the computational feasibility of the operator.

This paper is structured as follows. Section 2 introduces the preliminaries for our work and the refinement operator is presented in Section 3. There, we prove its ideality and describe how it can be optimised to work efficiently and incorporate background knowledge. We evaluate the operator on real—world knowledge bases in Section 4. Related work is described in Section 5 and conclusions are drawn in Section 6.

References

;

	Author	volume	Date Value	title	type	journal	titleUrl	doi	note	year
2009 IdealDownwardRefinementintheELD	Jens Lehmann Christoph Haase			Ideal Downward Refinement in the EL Description Logic						2009