<XML><RECORDS><RECORD><REFERENCE_TYPE>10</REFERENCE_TYPE><REFNUM>6938</REFNUM><AUTHORS><AUTHOR>Klampanos,I.A.</AUTHOR><AUTHOR>Jose,J.M.</AUTHOR></AUTHORS><YEAR>2003</YEAR><TITLE>Information Retrieval Over Semi-Collaborating Peer-to-Peer Overlay Networks</TITLE><PLACE_PUBLISHED>DCS Tech Report</PLACE_PUBLISHED><PUBLISHER>Dept of Computing Science, University of Glasgow</PUBLISHER><PAGES>13</PAGES><ISBN>TR-2003-139</ISBN><LABEL>Klampanos:2003:6938</LABEL><ABSTRACT>In this paper we study Information Retrieval (IR) over semi-collaborating Peer-to-Peer(P2P) networks. By the term semi-collaborating we mean networks where, although peers have to collaborate in order to achieve overall effectiveness, they do not have to share any proprietary information with the rest of the network, nor they have to be consistent with respect to the retrieval systems they use. For IR in particular, the potential of widely distributed information pools and its effortless access irrespectively of the underlying networking protocols, operating systems or devices, is utterly revolutionary and overwhelming. However, various limitations of existing systems have been identified with perhaps the most important one being that of the successful location of relevant information sources and the efficient query routing in large, highly distributed P2P networks. In this paper we propose a clustering-based architecture for IR over P2P networks in order to solve the aforementioned problem. We also research the usefulness of a simplified version of Dempster-Shafer (D-S) theory of evidence combination for the re-ranking of results in the network. Finally, we present the evaluation results we obtained through simulation in terms of the IR standard precision and recall measures. </ABSTRACT></RECORD></RECORDS></XML>