Context-Sensitive Ontology Matching in Electronic Business

Abstract

In today’s global economy, electronic business has offered great advantages to enhance the capabilities of traditional businesses. In order to satisfy the imposed requirement for businesses to coordinate with each other, electronic business partners are chosen to be represented by service agents. These agents need to understand each others’ service descriptions before successful coordination happens. Ontologies developed by service providers to describe their service can render help in this regard. Unfortunately, due to the heterogeneity implicit in independently designed ontologies, distributed agents are bound to face semantic mismatches and/or misunderstandings. This chapter introduces an innovative algorithm, Context-Sensitive Matching, to reconcile heterogeneous ontologies. This algorithm takes into consideration contextual information, via inference through a formal, robust statistical model based on confidence interval. In addition, an Artificial Neural Network is utilized to learning weights for different semantic aspects. At last, an agglomerative clustering algorithm is adopted to generate the final matching results.