Mutual Enhancement of Schema Mapping and Data Mapping

Schema mapping and data mapping have been two topics widely studied in traditional database related communities. Recently, with the rising interest in the semantical web, the tasks of integrating heterogeneous information sources, both in a schema and a data instance level, are becoming of more practical importance. Current efforts at resolving these two mapping tasks, namely, schema mapping and data mapping, have been carried out separately. In this paper, we propose a new method, which we believe, by simultaneously attacking these two tasks, will achieve a kind of mutual enhancement between them. By applying our method to a movie-hunting scenario, we show the precision and recall are both quite high. our method works well when dealing with numerical-valued attributes. We also show that this method is especially appealing from a semantic web-oriented perspective, given its assets of being relatively lightweight and easy to extend.