Expert search is mostly used in the areas of academic groups, organizations. The general expert search problem which is observed is searching experts on the web where lot of web pages and experts names are considered. It has two demanding issues namely 1) The quality and noisy content of web pages. 2) The names of the experts are usually unclear. We propose to influence the huge amount of coexisting information to access relevance and reputation of a person name for a query. The coexisting structure is modeled using a hypergraph, on which a heat dissipation based ranking algorithm is proposed. The keywords in the query are concerned as sources of heat, and name of the person which is closely related with the query will receive most of the heat, and ranked high. Our algorithms are efficient for retrieving experts and surpass baseline algorithms notably. This work would be noticed as one step toward addressing the problem in searching experts from the web.
展开▼
