Study On Recommendation Algorithm For Papers Based On Markov Chain And Node Centrality

With the rapid development of network technology, the number of academic papers is showing a trend of explosive growth in the field of scientific research. The overload of scientific research information, which also indirectly causes the waste of scientific research, makes the scientific researchers spend more energy accessing to resources. The active recommendation technology for papers is an effective way to solve this problem. However, there are some problems in the existing active recommendation technology for papers, for example the similarity of the paper, which is not easy to be quantified. At present, the social network has gradually become the mainstream of information recommendation and sharing. And more and more researchers use social networks to share academic resources. The recommendation technology for papers based on social network has become a new research hotspot. Therefore, it is the key problem to be solved in this research that how to simplify the calculation of the similarity of the papers and introduce the social network analysis into the active recommendation technology for papers.Firstly, based on the Markov Chain theory, a new thesis similarity calculation model is proposed. This model uses Markov chain to quantify the process of users’ browsing papers. And the state transition probability of the adjacent two times is regarded as the standard of measuring the similarity of the papers. Secondly, based on the idea of collaborative filtering, the social network technology is introduced into the construction of the similarity calculation model. And the users’ similarity calculation model is established by a series of process, which includes acquiring the state transition probability of all friends in the circle of friends, calculating the weight in consideration of the status of friends, weighting sum of the state transition probability. Thirdly, based on the users’ thesis similarity calculation model, a recommendation algorithm based on Markov chain and node centrality is designed. Finally, with the proposed algorithm as the core, this study designs and implements a paper recommendation prototype system. In addition, this study selects two indicators, precision rate and recall rate, to evaluate the effectiveness of this paper recommendation algorithm. And this study compares and analysis the recommendation effects of the recommendation algorithm based on VSM similarity and proposed in this paper. Experiments show that the recommendation algorithm proposed in this paper has higher recommendation accuracy, and the paper recommendation system has higher customer satisfaction.The recommendation algorithm proposed in this paper can be applied to the scientific research, and it can also be applied to the learning resources recommendation, the recommendation of the commercial and financial services in the financial industry. Therefore, the algorithm has wide application prospect.