Knowledge Vein-based Recommendation Of Academic Papers

With the rapid development of the Internet and the increase of ways to access information,the problem of information overload has become more and more serious.There are so many academic papers in the field of scientific research.People pay more and more energy to get interesting things from variety academic resources.The problem of information overload should be solved as soon as possible.Recommender systems are effective methods to solve this problem at present.In order to optimize the precision of academic recommendation approaches,this thesis proposes a recommendation approach based on Knowledge vein.This method calculates user interest based on knowledge vein which constructed by keywords and semantic relationship.This knowledge vein contains three kind of semantic relationships which are synonymy,hyponymy and co-occurrence relation.Then we make keywords from user's papers as the identity of their interest and the vector of keywords represents users and papers.Finally,the similarity between user and paper is measured by the semantic similarity of keywords which is calculated by the semantic distance in knowledge vein.After all these processing,we finish the recommendation of academic papers for users.This method uses the semantic association between keywords to recommend papers for users.This method takes it into account that the semantic information between keywords in one paper and the potential links between papers.Experimental results show that precision and recall of the recommendation has improved significantly.