Research On Dissertation Recommendation Algorithms Based On Citation Network

The emergence and popularization of the Internet has brought the huge amounts of data, which makes it difficult for users to find the most in need of their own from in the face of massive information, so information brings information utilization rate rapidly decreased, which is so called the information overload problem. Recommendation system in this context and produces, it is mainly depending on the interest and the user’s characteristics, recommended the most possible information of interest to the user or commodities, so as to solve the problem of the information overload. Recommendation system is more intelligent than the search engine, which is based on data mining.Recommended system in recent years has become a hot research in academia, because it is an era of information explosion, recommendation system can solve the information overload problem, has now been widely used in many fields, and has made in the industrial sector many successful applications. Much of the information (user metadata, the metadata item, collaborative similarity relationship between the user and the network feedback information, etc.) are useful for the recommended information service system. Traditional recommendation methods, including content-based recommendation, collaborative filtering, and recommendations are based on recommendations and hybrid social networks, which are respectively, for certain types of information to recommend a separate analysis, the effect will not get a better recommendation. This article is mainly for scientific research in the field, hoping to provide a unified recommendation model, the organic integration of all types of information together to improve recommendation results.Heterogeneous information network is a way to describe the multi-source, method of heterogeneous information can be effectively integrated into a variety of information into a model, this paper is the use of citations in this heterogeneous network information network as a recommended model to extract the author, a variety of useful information, such as research and as a recommended text similarity based recommender system in order to achieve the literature, the researcher recommended to achieve academic purposes.