Efficient Federated Debugging of Lightweight Ontologies

Abstract

In the last years ontologies have been applied increasingly as a conceptual view facilitating the federation of numerous data sources using different access methods and data schemes. Approaches such as ontology-based data integration (OBDI) are aimed at this purpose. According to these approaches, queries formulated in an ontology describing the knowledge domain as a whole are translated into queries formulated in vocabularies of integrated data sources. In such integrative environments the increasing number of heterogeneous data sources increases the risk of inconsistencies. These inconsistencies become a serious obstacle for leveraging the full potential of approaches like OBDI since inconsistencies can be hardly identified by existing reasoning algorithms, which mostly have been developed for processing of locally available knowledge bases. In this paper we present an alternative approach for efficient federated debugging. Our solution relies on the generation of so called clash queries that are evaluated over all integrated data sources. We further explain how these queries can be used for pinpointing those assertions that cause inconsistencies and discuss finally some experimental evaluation results of our implementation.

We refer the interested reader to our extended version of this paper ( http://www.researchgate.net/publication/263051384_Efficient_Federated_ Debugging_of_Lightweight_Ontologies ) comprising some additional parts like an introduction to inconsistencies in description logics, conjunctive queries and DL-Lite, a complete clash type definition, a complete list of derived clash queries, an example for clash query federation, the proof of Proposition 1, and a section about related work.