A mathematical model for improving retrieval performance in collaborative information retrieval

The accuracy of ad-hoc information retrieval (IR) systems has plateaued in the last few years. We are working on so-called collaborative information retrieval (CIR) systems which unobtrusively learn from their users’ search processes. As a first step towards techniques, we focus on a restricted setting in CIR in which only old queries and correct answer documents to these queries are available for improving on a new query. For this restricted setting we propose two initial approaches, called DTW and QTW, for reweighting the document-resp. the query-terms. We then combine these approaches with the pseudo relevance feedback method. The approaches are evaluated experimentally on standard IR test collections. It turns out that in particular the hybrid approaches with pseudo relevance feedback give promising results. A bigger advantage of the proposed approaches is expected in real word test scenarios in …