Research On Knowledge Recommendation In Academic Virtual Communities Based On UGC Mining

In the age of internet, the development of information technology bring the convenient of learning, living and working for human.The way for people to acquire knowledge greatly expanded by network, virtual community is one of this kind.Virtual community provide a platform for user to exchange information.With the subdivision of the content for information communicate, academic exchanges move to the internet to be a trend.The scholars pay more attention to use the network to make up the lack of academic exchanges in reality. Academic virtual communities attracted more and more scholars to participate for its fast and openness, which can not only meets the demand of academic exchange for scholars under the subdivision of information communication, but also become an expansion and extension of traditional academic exchanges on network, and to be an important platform for professionals and academic workers to share information and knowledge. Knowledge exchange and sharing is the basis of the existence and development of academic virtual communities. The geometric growth of user interaction and knowledge amount in academic virtual communities, the knowledge demanders how to interact with academic scholars, the scholars how to obtain knowledge efficiently, accurately and quickly, at the mean while, how to find the learners who have the same interest, and how to enhance the degree of knowledge exchange and sharing, are all questions should consider for the research of academic virtual communities in recently.This thesis is based on the research of academic virtual communities, user generated content, web data mining and knowledge recommendation.As define its concept respectively, this paper put forward the concept of academic UGC, the analysis the different features of academic UGC from the other UGC in the content, users and link dimension, using the Web content mining and structure mining in the mining of academic UGC, then use the semantic-based k-means dustering to achieve academic UGC ontent mining, established the explicit knowledge of academic recommendation model in academic virtual communities; optimized user distance calculation method in academic virtual communities, derived improved algorithms PageRank2.0to use in academic UGC link mining, established tacit knowledge recommendation model, recommended other users who have the same research interests to users.Finally, use the UChome website building tools to build an academic virtual community have the theme of "data mining",were analyzed23user who created250blogs. Analysised the text and links of those250blogs and then to do the text dustering and user authority calculation. Academic UGC preprocessed by text preprocessing and feature representation, then use the blog feature representation set to calculate the text similarity and come to the similar matrix, at last use the k-means algorithm to cluster similarity matrix; analysis the into chain links in academic UGC, give the chain link matrix to calculate the value of each user’s authority, then recommend the authority of relevant areas after the clustering of academic UGC, this is the way of tacit knowledge recommendation. Such knowledge and knowledge sources(academic authority) of knowledge recommendation, can promote "learners" to exchange and share knowledge, and spiral the stock of knowledge in entire academic community.