Research On The Method Of Constructing Expert Relationship Network Based On Theme

Experts retrieval is one application basic research in the field of information processing, which has important applications in science, business management and other occasions, it’s the most effective means of obtaining expert information currently. Experts is not alone in complex social networks, they contacted together through the complex relationship, constitute the field have the same relationship expert knowledge sharing network, this network of relationships is an important resource to support retrieval experts, the reliability of a direct impact on retrieval expert accuracy. Because experts who have relationships tend to have the same field of knowledge sharing, and the sharing of knowledge in the field can be used to characterize by the expert implicit theme. Therefore, we focus on extracting the theme of expert implicit, focus on extracting the expert relationship of implicit theme and building networks of experts to carry out a series of issues such as research and discussion, several major achievements made in the following areas:(1) Proposed a expert model building methods of integrating specialists cooperative network probability model. The method focus on analysising specialists cooperative network, make full use of expert cooperation between adjacent nodes in the network have similar themes distributed features in the probabilistic model of Gibbs sampling topic among experts topic distribution constraints, thereby enhancing expert topics clustering effect. Making expert themes extracted experimen based on the NIPS dataset and results show that the fusion experts cooperation experts and probabilistic topic model network model can effectively extract the theme topic experts, expert cooperation network of experts to extract themes have a supporting role, the model has a more strong promotional.(2) Proposed a building methods of extracting expert relationship based on Markov. Thinking that character of traditional entity-relationship extraction method can not be fully applicable to the experts relation extraction task,this method to analysis the relationship between friends, colleagues and guidance with the help of expert resource, and integration specialists relationship characteristics and associated features thematic relationships to construct experts relationship extraction model based on Markov network. Experimental results show that the proposed expert relationship extraction model based on Markov network and achieved good results in the recognition expert friend relationships between colleagues and mentoring relationships, relationship experts topic for experts to identify good supporting role.(3) Proposed a method to build an expert network of relationships based on the strategy of random walk. Firstly, according to the extracted relations experts to build expert relationship matrix, then by means of a random walk strategy thinking, would not some simple characterization of experts to map the relationship between organic combination, in order to build a complex network of relationships expert. The experimental results show the effectiveness of construction method based on random walk strategy proposed expert networks.(4) Design and implementation of expert networks to build a prototype system built to further study the relationship between the network of experts to provide a platform for experts to retrieve provide important support.