Optimizing Queries for Heterogeneous Information Sources

Abstract

Integration of heterogeneous information sources requires building an infrastructure for accessing several different information sources together. One task of the integration is to provide a uniform interface to the different information sources (databases, directory servers, web services, web sites, XML files). The other task is the semantic integration: the meaning of the stored data is also different in the different sources.

In the SILK [1] system (Semantic Integration with Knowledge and Logic), semantic integration is carried out by providing a high-level model and the mappings between the high-level model and the models of the sources.When executing a query of the high-level model, the query has to be transformed to queries of the sources and to the code performing the semantic transformation of the data. The component which translates a high-level query to a low level query is the Query Planner. The Query Planner builds the set of low-level queries, and a further component is needed to create an efficient query plan from it. This component in the SILK system is called the Query Optimizer.