Détection de liens d'identité erronés en utilisant la détection de communautés dans les graphes d'identité

Abstract

Different studies have observed that the semantic web identity predicate owl:SameAs is sometimes used incorrectly. In this paper, we show how network metrics such as the community structure of the owl:SameAs graph can be used in order to detect such possibly erroneous statements. One benefit of the here presented approach is that it can be applied to the network of owl:SameAs links, and does not rely on any additional knowledge. We evaluate our approach on 558M owl:SameAs statements scraped from the LOD cloud. This evaluation shows the ability of our approach to scale, and its efficiency in detecting erroneous identity links.