Affiliations: Institute AIFB, University of Karlsruhe, Germany.
Abstract: In this paper we presented a framework for query refinement which is
driven by user's information needs. Based on the analyses of the real web IR
case studies we model the query refinement process as the process of decreasing
query ambiguity with respect to the user's need. In the refinement process a
user is provided with the query's neighborhood and the guidelines for changing
the ambiguities of the query. The approach assumes the existence of a
controlled vocabulary, which is used for describing the content of information
resources. We model the queries (query-answering pairs), ambiguities and
refinements formally. The partial ordering of the query-answering pairs in the
searching space represents the mathematical lattice structure, so that formal
concept analysis is used as the mathematical backbone of the query refinement
process. We present a case study related to the searching in MEDLINE, which
shows the benefits of using our approach.
Keywords: Query by navigation, query refinement, ontology, information need