Credibility Measurement Of Network Information Based On Domain Knowledge Graph

The credibility of network information is related to the effectiveness of decision making.Network information in social media,mobile Internet and big data environment has many new characteristics,such as users' active participation,information multi source heterogeneity and massive dynamic.In the new network environment,the process of information interaction is becoming more and more complex,the authenticity or credibility of information is getting more and more attention,especially for the more and more research on the evaluation and measurement of the reliability of the network information.This paper tries to use knowledge map as a tool to measure the reliability of network information by constructing knowledge graph,so as to understand and feel the specific process of measuring the reliability of network information.Knowledge graph is a form of atlas organization.It associates various entities through semantic association,and displays the knowledge base based on Semantic Web.It is important to extract the relationship and show high correlation and high structural results.Semantic web is for computer understanding of human meaning,so graph structure provides a good support for reasoning.Knowledge graph extracts and integrates structured and unstructured data through data and embodies the idea of data governance and semantic connection,which is beneficial to the use and migration of large-scale data.Knowledge graph,as a new knowledge organization and retrieval technology in the big data age in the past two years,has been gradually embody the advantages of knowledge organization and display,and has attracted much attention from many fields,and its application prospect is very wide.However,the current development of knowledge graph is still in the primary stage,facing many challenges and problems,such as the automatic extension of knowledge base,the processing of heterogeneous knowledge,the learning of reasoning rules and so on.Although related research has put forward the treatment and improvement models for different problems,we need to further study them.Based on the expert user experience and the popular user experience in social media platform,this paper studies the construction of knowledge graph from three aspects: text clustering,social network analysis and text classification.In view of the difficulties in the construction of knowledge map and the shortcomings of related research,this paper makes some discussions and researches,mainly including the following aspects:(1)This paper attempts to build domain knowledge graph based on expert user experience in the field,as a reference standard in the field.The data originate from the different expert experience knowledge of several professional websites,which can not only reflect the cross validation of different expert experiences,but also make the dimension of the domain knowledge more comprehensive,and prevent the one-sided nature of the different tendencies of the single website or expert experience.Based on hierarchical clustering,the knowledge graph of homogeneous network clustering is constructed,and the similarity degree and association relationship between the same type nodes are revealed;Based on social network analysis,a knowledge graph of heterogeneous network clustering is constructed,which reveals the relationship between two different types of nodes,and makes up for the shortcomings of the traditional multidimensional scale analysis method.(2)Based on expert user experience,we build knowledge graph based on mass user experience in social media platform.Because the public users lack a systematic and comprehensive knowledge of domain knowledge,resulting in very sparse and fragmented interactive data,it is necessary to take expert experience as a reference standard.The user interaction data in the social media platform is Natural Language Processing,and the expert lexicon is used as the Chinese word segmentation dictionary,which makes the processing result more standardized.Also,based on hierarchical clustering and social network analysis,the knowledge graph of homogeneity network clustering and heterogeneous network clustering is constructed to reveal the relationship between nodes and facilitate the analysis and comparison between the two.(3)We compare and analyze the mass user experience and expert user experience in the social media platform to measure the credibility of the mass user experience.The similarity between words is calculated based on KL divergence.The higher the similarity is,the more credible it is.Then the KL values are arranged from small to large,the threshold is set and the threshold setting is evaluated by cross validation,and the threshold setting of the highest index value is selected as the most robust classification threshold to ensure the classification results.It is important to note that the confidence measure given in this paper is only a possibility of the credibility of the content of the network information,but only a relative standard is given within the scope of the community research,and it is not an absolutestatement that the scope of the field is not necessarily effective.Therefore,the related researches in the future need further improvement.