Research On Visualization And Core Document Recommendation Technology In Citation Network

With more and more rapid growing of scientific literatures, when researchers are facing the returned results from search engines, how to pick out relevant and valuable literature with their own field of research has currently become one of the key research technologies. However, most of the traditional citation analysis methods only use the citation counts and the factor of journal to evaluate documents, and the existing papers visualization layout algorithms cannot reflect the importance of papers. Therefore, combining with the PageRank technology and network visualization technology can be used to find out valuable documents quickly and also can dig out hidden information from the literature network.Based on the summation of citation behaviors, evaluating methods and literature network visualization technology, the key problems in evaluating methods and literature network visualization technology are analyzed in this thesis. The algorithm of evaluating method based on PageRank and core documents visualization are studied in focus.Firstly, as most of evaluating methods cannot analyze the value of documents objectively and accurately, a new evaluating algorithm is proposed. At the same time, the importance of document is defined. It is improved based on the PageRank algorithm, and evaluates the value of documents by analyzing the paper's inherent value and the value passed from the behavior of citation, and the factor of personalized interaction is added too. The experimental results show that this method can not only solve the results tending to old documents and theme drifting by traditional evaluating methods, but also can let researchers choose the valuable literature according to their own needs.Secondly, since the present studies of literature network visualization are simple, citation network and co-author network are separated to study, and lack of research on core documents visualization, a two-layer visualization layout model based on citation and co-author network is built, and a new intuitively display method for results of evaluating by using core documents visualization is proposed. The experimental results show that this model and the method can greatly improve the literature network visual aesthetics, and researchers can accurately and objectively grasp the trend of development. Moreover, the interactive feature enables to get more detailed information.Finally, the evaluating method and two-layer network model are integrated, and an interactive scientific literature recommendation system is designed and developed. This system can basically meet the needs to find valuable literature for researchers.