Research Of High-quality Personalized Paper Recommender System

Personalized recommendation system is a new type and user-centered platform to access knowledge. At present, due to the data that building user's interests is not authentic and enough; the quality of the waiting recommended paper can't be guaranteed; the system can't find new interest for user; most of the recommendation systems are blind and inefficient. To solve these problems, we propose the high-quality personalized paper recommender system, there are three improvement in this system.First, we propose a method to score the research papers, solving the problem that lacking the evaluation of paper in paper recommendation, this method considers comprehensive factors that affecting the quality of the paper, including Impact-Factor, the title of the author, the fund of the paper, the frequency of quotation. So it can more objective, more fair, can reflect the quality of the paper truly.Second, building user's interests from the paper that they published, saving the valuable time of the researcher, guarantee the quality of the interesting data, that is decided by the feature that researcher's interesting is steadily relatively.Last, we propose an algorithm to find new direction of interest in the user's research field. In the past, the content-based recommendation system can only recommend similar paper to the users, can't find new interest to users, this system can do it.Experiments show this system can solve these problems, and improve the quality of the paper recommendation system.