A Study of the Dependencies in Expert Finding

Expert finding, the task of identifying experts on a given topic, has received much attention in the last few years. Existing methods try to model the dependencies between the candidate experts and the query terms, which have been proven to be effective. However, to the best of our knowledge, no work has been done on studying the effectiveness of another kind of dependencies, namely the dependencies between the query terms, for expert finding. In this paper, we exploit the dependencies between the query terms for expert finding and propose a language modeling-based approach which combines the two kinds of dependencies within a unified framework. Specifically, we first propose a window-based method to model the dependencies between query terms. And then we combine it with the co-occurrence model which is designed to model the dependencies between the candidate experts and the query terms. Experiment results demonstrate the effectiveness of the window-based method and show that a combination of both kinds of dependencies can improve the performance significantly.