Recommendation Algorithm Based on Topic Utility for Academic Papers

To solve paper recommendation problem,an academic paper recommendation algorithm based on topic utility is proposed. This approach uses latent Dirichlet allocation(LDA) model to build the model of candidate papers and users' published papers,and then the topic sets with high utility are mined. The similarity between the user interest and the candidate papers is calculated according to the distribution of the high utility topics. Finally,the valuable papers are recommended to the users.Experimental results show that this method is effective,and it can get higher precision and recall than the algorithm based on apriori. Meanwhile,this method can meet the user demand for both the quality and the personalization.