Information source selection for resource constrained environments

Abstract

Distributed information retrieval has pressing scalability concerns due to the growing number of independent sources of on-line data and the emerging applications. A promising solution to distributed retrieval is metasearching, which dispatches a user's query to multiple sources and gathers the results into a single result set. An important component of metasearching is selecting the set of information sources most likely to provide relevant documents. Recent research has focused on how to obtain statistics for the selection task. In this paper we discuss different information source selection approaches and their applicability for resource-constrained sensor network applications.