Evaluating Matching Algorithms: the Monotonicity Principle

Ateret Anaby-Tavor, Avigdor Gal, Alberto Trombetta

Abstract: In this paper we present the monotonicity principle, a sufficient condition to ensure that exact mapping, a mapping as would be performed by a human observer, is ranked close to the best mapping, as generated automatically by a matching algorithm. The research is motivated by the introduction of the semantic Web vision and the shift towards machine understandable Web resources. We support the importance of the monotonicity principle by empirical analysis of a matching algorithm, showing that algorithms that obey this principle rank the exact mapping close to the best mapping. keywords: Ontology matching, Novel integration architectures