Expert Retrieval Community-oriented Mining Research

With the rapid development of Internet, people continue to publish some information on the Internet, Internet has become a huge source of information. This greatly improves people's daily life and exchange of information has greatly been changed. People can even get the important information emerged in this world one day. In addition, because network resources are growing, it also brings some negative effects; people tend to be flooded by the explosive information and do not know what the real information they need is. How to effectively use the network on the Web and found useful information becomes a main research content of data mining. Traditional data mining can only provide people with some basic text information, but with the rapid development of human information, people find that simple document search are no longer able to meet the needs of users.In recent years, expertise retrieval as a rising research hot spot has attracted much attentions in information retrieval and knowledge discovery. It aims at finding professional expertise in specific area through a series of measures to provide decision support for scientific research and business management. As a new task of expertise retrieval, finding research communities or teams for scientific guidance and research cooperation has become more and more important.However, the existing community discovery algorithms only consider graph structure, without considering the context, such as knowledge characteristics. Therefore, detecting research community cannot be simply addressed by direct application of existing methods. This thesis proposes a hierarchical discovery strategy which rapidly locates the core of the research community, and then incrementally extends the community. Especially, as expanding local community, it selects a node considering both its connection strength and expertise divergence to the candidate community, to prevent intellectually irrelevant nodes to spill-in to the current community. The experiments on ACL Anthology Network show our method is effective.