Advances in Ontology Matching Avigdor Gal Technion – Israel Institute of Technology avigal@ie.technion.ac.il Pavel Shvaiko University of Trento, Povo, Trento, Italy pavel@dit.unitn.it Abstract Matching of concepts describing the meaning of data in heterogeneous distributed information sources, such as database schemas and other meta- data models, grouped here under the heading of an ontology, is one of the basic operations of semantic heterogeneity reconciliation. The aim of this chapter is to motivate the need for ontology matching, introduce the ba- sics of ontology matching, and then discuss several promising themes in the area as reflected in recent research works. In particular, we focus on such themes as uncertainty in ontology matching, matching ensembles, and matcher self-tuning. Finally, we outline some important directions for future research. 1 Introduction Matching of concepts describing the meaning of data in heterogeneous dis- tributed information sources (e.g., database schemas, XML DTDs, HTML form tags) is one of the basic operations of semantic heterogeneity reconciliation. Due to the cognitive complexity of this matching process [18], it has tradition- ally been performed by human experts, such as web designers, database ana- lysts, and even lay users, depending on the context of the application [79, 47]. For obvious reasons, manual concept reconciliation in dynamic environments such as the web (with or without computer-aided tools) is inefficient to the point of being infeasible, and so cannot provide a general solution for semantic reconciliation. The move from manual to semi-automatic matching has there- fore been justified in the literature using arguments of scalability, especially for matching between large schemas [45], and by the need to speed-up the matching process. Researchers also argue for moving to fully-automatic, that is, unsuper- vised, schema matching in settings where a human expert is absent from the decision process. In particular, such situations characterize numerous emerging applications, such as agent communication, semantic web service composition, 1