DIPROM: DIstance PROportional Matcher Exploiting Neighbor-levels and Related Terms

Abstract

Schema matching has the task to find semantically related elements in the input schemas. Many automated schema matchers have been proposed, but none of the solutions performs consistently without mismatches. In this paper, we propose a new schema matcher which contains enhanced matching techniques in its components to improve accuracy. Especially, this approach exploits the so called related term sets to provide a more accurate matching, quickly. Since the related terms sets are rarely provided, a process is also proposed to extract related term sets from schema descriptions. Another specialty of our schema matcher is the definition of entity neighbor-levels. This technique is applied to evaluate the neighbor-level similarities and include these values in the assessment of entity relatedness. Our approach has been shown performing reliably and has been compared with other schema matchers both as a stand-alone hybrid schema matcher and component-wise.