UNIVERSITY OF TRENTO DIPARTIMENTO DI INGEGNERIA E SCIENZA DELL���INFORMAZIONE 38050 Povo ��� Trento (Italy), Via Sommarive 14 http://www.disi.unitn.it APPROXIMATE STRUCTURE-PRESERVING SEMANTIC MATCHING Fausto Giunchiglia, Fiona McNeill, Mikalai Yatskevich, Juan Pane, Paolo Besana and Pavel Shvaiko August 2008 Technical Report # DISI-08-039 Also: accepted to the 7th International Conference on Ontologies, DataBases, and Applications of Semantics (ODBASE 2008), Monterrey, Mexico, Nov 11 - 13, 2008.

.

Approximate structure-preserving semantic matching Fausto Giunchiglia1, Fiona McNeill2, Mikalai Yatskevich1, Juan Pane1, Paolo Besana2, Pavel Shvaiko3 1 University of Trento, Povo, Trento, Italy, {fausto|yatskevi|pane|pavel|}@dit.unitn.it 2 University of Edinburgh, Scotland, f.j.mcneill@ed.ac.uk|p.besana@sms.ed.ac.uk 3 TasLab, Informatica Trentina, Italy, Pavel.Shvaiko@infotn.it Abstract. Typical ontology matching applications, such as ontology integration, focus on the computation of correspondences holding between the nodes of two graph-like structures, e.g., between concepts in two ontologies. However, for ap- plications such as web service integration, we need to establish whether full graph structures correspond to one another globally, preserving certain structural prop- erties of the graphs being considered. The goal of this paper is to provide a new matching operation, called structure-preserving semantic matching. This opera- tion takes two graph-like structures and produces a set of correspondences, (i) still preserving a set of structural properties of the graphs being matched, (ii) only in the case if the graphs are globally similar to one another. Our approach is based on a formal theory of abstraction and on a tree edit distance measure. We have evaluated our solution in various settings. Empirical results show the efficiency and effectiveness of our approach. 1 Introduction Ontology matching is a critical operation in many applications, such as Artificial In- telligence, the Semantic Web and e-commerce. It takes two graph-like structures, for instance, lightweight ontologies [9], and produces an alignment, that is, a set of cor- respondences, between the nodes of those graphs that correspond semantically to one another [6]. Many varied solutions of matching have been proposed so far see [6,29,24] for re- cent surveys4. In this paper we introduce a particular type of matching, namely Structure- preserving semantic matching (SPSM). In contrast to conventional ontology matching, which aims to match single words through considering their position in hierarchical ontologies, structure-preserving semantics matching aims to match complex, structured terms. These terms are not structured according to their semantics, as terms are in an ontology, but are structured to express relationships: in the case of our approach, first- order relationships. This structure-preserving matching is therefore a two-step process, the first step of which is to match individual words within the terms through tech- niques used for conventional ontology matching, and the second - and novel - step of 4 See, http://www.ontologymatching.org for a complete information on the topic.