Using Plausible Inferences and Dempster-Shafer Theory of Evidence for Adaptive Information Filtering
Farhad Orumchian, Babak Nadjar Araabi, Elham Ashoori · 2002
Abstract: This paper describes a set of experiments investigating the use of Dempster-Shafer (DS) theory of evidence in combining the evidences provided by plausible inferences in Information Retrieval. PLIR is an experimental information retrieval system based on the theory of plausible reasoning[1] which tries to simulate a reference librarians reasoning while trying to retrieve documents in response to queries. Often a document can be retrieved by several plausible inferences. Each inference constitutes of an uncertain evidence of relevance. In these experiments, DS theory is used in several ways to combine these evidences in PLIR system. The application of DS theory and weighting inferences based on their overall usefulness in retrieval has improved the quality of ranking of PLIR system. We use this conclusion in our future research in the application of plausible inferences and DS theory in adaptive information filtering.