by Pavel Serdyukov (University of Twente)
Expert finding is one of the most rapidly developing IR tasks and a popular research domain. The opportunity of search for knowledgeable people in the scope of an organization or world-wide is a feature which makes modern Enterprise search systems commercially successful and socially demanded. A number of efficient expert finding approaches is proposed recently. Although, most of them are based on reasonably defined measures of expertness, they still use rather unrealistic and oversimplified principles. In our research we try to avoid these limitations and come up with models that go beyond the assumptions used in state-of-the-art expert finding methods. In one approach, we suppose that while the probability of co-occurrence of the person and query terms in top ranked documents is a reasonable measure of expertise, the assumption of their independence is not quite adequate. Regarding persons as generators of terms in top documents, we build a model of co-occurrence that appears to be more effective than the model assuming their independence. In our another approach, we rely on a model of manual search for an expert, but in contrast to existing approaches, we do not assume that the user stops after the first step of moving from selected document to the found candidate expert. We represent the search for expertise as a finite or infinite random process of consulting with both possible sources of knowledge: documents and people. Both approaches and the current state of expert finding research raises a lot of questions which we suggest for discussion in this paper.