Improving Document Search by Finding Domain Experts

Domain experts are important clues to search for relevant documents in digital library information system. In this paper, we present a method to improve the document search by finding domain experts. Our method aims at exploiting the domain expertise as a rich source of evidence for document search. In particular, we propose (i) a graph-based model of the domain experts, (ii) an algorithm to find the experts based on this model, and (iii) a novel probabilistic framework to incorporate the expert information into document search. In the experiments, we find that the conditional probability of document relevance on domain expert, i.e. P(D|e, q), follows a geometric distribution. The Experimental evaluation also demonstrates that the expert-based document search greatly increase the retrieval effectiveness.