Analysis And Prediction Of Research Article Impact Based On Social Media

How to excavate the valuable article quickly and accurately is a critical goal of the research of the paper impact,and the citation has become widely used to evaluate the impact of a paper.However,we cannot use the citation to evaluate the impact of a recent paper,because it needs a long time to get enough citations.In this paper,we use the social media data to analyze and predict article impact.We use real-time data from social media to evaluate the influence of recent papers.Firstly,we analyze the development trend of coverage and mention rate of different papers published in different years on Twitter and Facebook.Then we systematically study the distribution of academic papers in social media to ensure social media data is available enough to assess the influence of the paper.Secondly,we analyze the relationship between paper citation and social media citation,and explore whether there is a definite correlation between them.The study finds that the paper citation and social media citation still reflect the impact of the paper in different ways,and the relationship between them is influenced by many factors.Based on the above research,we introduce a new prediction model(P-K model)to predict the paper citation after 3 years of publication to evaluate the impact of recent papers.In P-K model,we considered a series of objective features of papers,especially the keywords and use social media citations to predict paper citations.We conducted extensive experiments on the Nature journal data set to evaluate the performance of P-K model and regression models in 4 disciplines.The results demonstrate that,P-K model performs best on citation impact prediction and the results are closer to the real paper citation than the regression model on 4 disciplines.