The iSeM matchmaker: A flexible approach for adaptive hybrid semantic service selection

We present iSeM (intelligent Service Matchmaker), a precise hybrid and adaptive matchmaker for semantic Web services, which exploits functional service descriptions in terms of logical signature annotations as well as specifications of preconditions and effects. In particular, besides well-known strict logical matching filters and non-logic-based textual and structural signature matching, it adopts approximated reasoning based on logical concept abduction and contraction for the description logic subset SH with information-theoretic valuation for matching inputs and outputs. In addition, it uses a stateless logical specification matching approach, which applies the incomplete but decidable@q-subsumption algorithm for preconditions and effects. The optimal aggregation strategy of all those aspects is learned off-line by means of a binary SVM-based service relevance classifier in combination with evidential coherence-based pruning to improve ranking precision with respect to false classification of any such variant on its own. We demonstrate the additional benefit of the presented approximation and the adaptive hybrid combination by example and by presenting an experimental performance analysis.