Schema adaptive modeling and incremental matching for web interface

Abstract

There are so many web data hidden behind so-called deep web and can only be accessed through query interfaces, and the data volume is increasing. We often need to fill forms over alternative interfaces in the same domain to select the best product or service, such as buying a book across various online book websites to choose the most affordable one. Integrating these web interfaces in the same domain to an uniform interface is as a matter of course. One of the most important things for interface integration is interface schema matching. In this paper, we present a new structure and a corresponding algorithm for web interface schema modeling and matching. Using the new schema structure, we could not only easily handle two interfaces schema matching, but also handle incremental schema matching between an existing integrated interface and a new interface. We present a detailed experimental evaluation using UIUC Web Integration Repository dataset. The results show that our approach is effective: it obtains significantly higher accuracy for two schema matching and is more robust than previous techniques. For incremental schema matching, it also performs well.