Interpreting similarity measures: Bridging the gap between schema matching and data integration

It has been recognized in the literature that the process of schema matching is uncertain. Such uncertainty at the core of data integration needs to be managed correctly to avoid dire consequences. Traditionally, manual intervention was required to make local decisions at the schema matching level to reach a deteministic matching before the rest of the data integration system can use it. Recently, however, researchers have arguedfor moving tofully-automatic transition of schema matching results into other data integration activities. In this work we discuss what it takes to bridge the gap between automatic schema matching and data integration. We briefly present the modeling of schema matching as an uncertain process, review a sufficient condition for using matcher similarity measure as a measure of schema matching correctness andprovide a case study of data integration in Peer Database Management System to demonstrate the benefit of our proposed gap bridging technique.