Algorithm Design Of Network Articles Influence Assessment

In recent years,with the development of Internet services,the network and new media has become an important position of ideological confrontation,all kinds of cultures and values are communicate with each other on the Internet,people can tell others their ideas and access to more and more information than ever before,including speeches from someone or some enterprises.However,how to appraise the influence of their opinions and encourage them to speak correctly on the Internet is still a challenging topic.This paper firstly shows the research status in the world today,mainly introduced new and old ways to calculate the influence of blog post,micro-blog news,and other new medias appeared,such as "the method based on the H index" etc..Then this paper introduces the basic knowledge about the algorithm in this paper,including massive text similarity calculation algorithm and SimHash algorithm and Hamming distance calculation method,and also includes algorithms to analyse network links such as HITS algorithm,and the calculation of article authorities by using the classical PageRank algorithm;then we measure network influences from five big dimensions,and set up the author's social networks and the article network links,computing each small index under each big dimension.Then we get samples of article influences through manual scoring,and get the final algorithm to calculate influences of articles by back propagation neural network.It is proved that the experimental results are reasonable.The influences of articles calculated by the algorithm can represent the real influences of articles.A major difficulty of this algorithm is article's influence definition,we have a standard definition about high influence:the article has a large number of audiences and give the readers more impressions has a heigher influence,the article itself has nothing to do with influence.So we validate the calculation results of the algorithm by comparing with manual scoring results,people does not consider who the author is when scoring,when calculate the influence the algorithm considers the indexes about authors,and comparing the calculated results with the results of manual scoring.Experimental results show the effectiveness of the proposed algorithm.