Query rewriting for lightweight information integration

Light-weight information integration has to be able to effectively deal with settings where only partial mappings between the information sources to be integrated are defined. In our query rewriting approach, such partial mappings are used as the basis for defining several strategies to substitute elements in the query, and for introducing wildcards for relaxing the query when mappings are missing. The approach also supports the extension and refinement of the existing mappings based on the achieved query results and optional feedback of the user. This paper presents the proposed rewriting strategies, focusing on strategies exploiting ontological knowledge, and a comparison and evaluation of the strategies for real-world datasets.