Semantics-guided clustering of heterogeneous XML schemas

In this paper we illustrate an approach for clustering semantically heterogeneous XML Schemas. The proposed approach is driven by the semantics of the involved Schemas that is defined by means of the interschema properties existing among concepts represented therein; interschema properties taken into account by our approach are synonymies (indicating that two concepts have the same meaning), hyponymies (denoting that a concept has a more specific meaning than another one), and overlappings (indicating that two concepts are neither synonyms nor one hyponym of the other, but represent, to some extent, the same reality). An important feature of our approach consists of its capability of being integrated with almost all the clustering algorithms already proposed in the literature. Both a theoretical and an experimental analysis on the complexity of our approach are presented in the paper. They show that our approach is scalable and particularly suited in application contexts characterized by a great number and a large variety of XML Schemas.