Research On Credibility Evaluation Of Social Information On Social Media Based On Content Analysis

With the development of Internet technology,the threshold of media information release has been rapidly reduced.The emergence of Internet social media makes everyone become an information publisher instead of merely information receiver.The massive amount of information on the Internet has become a part of people's daily life.However,the characteristics of large amount,low release threshold and fast transmission speed also make the quality of social life information in social media decline significantly compared with traditional media.The uncontrollability and harmfulness of false information has become increasingly obvious.Some of the false information even has some real content and strong logic,which bring's great difficulty for users to distinguish such false information.In view of the frequent occurrence of false information in social media,scholars have conducted many studies on the credibility of social media.However,the current research methods are relatively isolated among various subjects.In the field of communication,the questionnaire and interview methods which are similar to traditional media credibility survey are usually performed.However,the shortage in the universality and quantity of samples are obvious.In the field of computer science,automatic identification is excessively emphasized.The neural network is used to judge fully automatically after obtaining a large amount of information.At present,the recognition of images is ideal but the performance of semantic recognition is poor due to the limitation of artificial intelligence,limiting the analysis of the credibility,in which text processing predominated.In this thesis,we combined the data acquisition methods in the field of computer science and the coding analysis in the field of information science,taking Weibo and Zhihu as examples to study the credibility evaluation of social life information in social media.The main contribution of this article is as follows:First,the python crawler is used to obtain the comments of users on the selected topics in Weibo and Zhihu platforms.Because users do not know that their speeches will be obtained in advance,they will make their own speeches.The original comments obtained by Python crawler which are related to credibility evaluation are selected as research samples.Secondly,based on the basic theories in the fields of information science,communication science and credibility research,this paper summarizes the existing research on the credibility evaluation index of social media information at home and abroad,and constructs the credibility evaluation index system of social life information of social media,which serves as the framework for subsequent research based on content analysis.Two postgraduates majoring in Library and information science form a coding group,and they code and sort out the comment samples respectively.In addition,the consistency of the coding process was tested,and the reliability and validity of the whole research process were tested,which verified the objectivity and effectiveness of the research process.Finally,this thesis summarizes and analyzes the coding results,analyzes the frequency of users' use of evaluation indicators and the differences of indicators under different topics,and sums up the characteristics of users' credibility evaluation of social life information of social media.Some suggestions were proposed to improve the credibility of information from the perspective of the information publishers,domain authority users,ordinary users,official institutions,social media platforms,laws and regulations,relevant functional departments.