Credibility Evaluation Method Of Social Media News Based On Multi-source Information Fusion

Social media has greatly improved the efficiency of news dissemination.However,since the news published on the social media platform is no longer verified by journalists or experts,fake news is rampant on the social media platform,which poses a great threat to the long-term stability of the society.Therefore,it has always been an important task for social media platforms and government agencies to detect and rectify fake news.However,the traditional manual identification method has been unable to adapt to the explosive growth of social media news,so automated methods are urgently needed to complete this task.In view of this demand,the automatic evaluation of social media news credibility has gradually become one of the research hotspots in academia and industry,and the key lies in distinguishing the authenticity of events described by social media news.However,most of the existing studies use a single report to evaluate the authenticity of news events,and the results obtained are difficult to be comprehensive and objective.In addition,although the researchers have made some progress and breakthroughs in the theory and algorithm of automatic evaluation of social media news credibility,the complete system implementation scheme is still relatively rare.To solve the above problems,we propose a Credibility Evaluation Framework of Social Media News Based on Multi-source Score Fusion(CEFN),which uses the multisource of social media news reports and integrates the credibility of multiple reporting sources to determine the credibility of news events.Therefore,CEFN's assessment of the credibility of social media news is more comprehensive and objective than that relying on single news report.In addition,CEFN fully considers the author's information and the article's information of news sources in the credibility evaluation of news sources.Compared with the evaluation method relying on a single feature,CEFN takes more comprehensive factors into account and the assessment results are more accurate.CEFN provides a complete implementation scheme for the automatic evaluation system of social media news credibility,which mainly includes three functional modules: multi-source news collection module,news source score evaluation module and multisource score fusion module.First of all,the multi-source news collection module collects and filters news articles from social media platform through a three-step method,which aims to collect more source data support for the evaluation of news credibility.Then,the news source score evaluation module scores the credibility of a single news source through two sets of feature sets(that is,the author-based feature set and the article-based feature set),and the score is used to describe the news source's support for the authenticity of the news.Finally,based on the DS evidence theory,the multi-source scoring fusion module fuses the credibility of multiple news sources and forms the final judgment result.In this paper,we use real social media news data to test the performance of CEFN,and the experimental results verify the effectiveness and advancement of CEFN in the credibility assessment of social media news.