Research On Business Reputation Evaluation Combined With Sentiment Analysis In Social Commerce

Under the mutual infiltration of social media and e-commerce,social commerce with both attributes are derived,and with the rapid spread of social media,social commerce has also developed rapidly.However,due to the increasing number of business in the social commerce platform,the online market trading environment has not been fully regulated,resulting in more and more prominent business reputation issues.In addition,transactions between business and consumers in the social commerce environment are becoming more frequent,the number of user review text is rising sharply,and consumers often have strong subjectivity and arbitrariness in their evaluations,resulting in user review texts with very complex emotional color.So it is very difficult for consumers to quickly find valuable information from these vast amounts of review text data,which can be used to identify the authenticity of the business reputation and make reasonable purchasing decisions.In order to solve the above problems,this paper is based on the short text of user review in the social commerce platform,using various text processing techniques to extract the business reputation dimension that consumers pay more attention to,establishing a complete evaluation index system of business reputation,and dividing business credit rating.Finally obtaining the comprehensive evaluation results of the business reputation.The main research contents are as follows:(1)The construction of business reputation dimension Firstly,the short text of social commerce review is preprocessed.For the new network words not included in the HNC knowledge base,this paper proposes a method for processing this kind of words according to the composition rule of HNC symbols and by means of network retrieval.The semantic similarity calculation is carried out by using the complete HNC symbol.Then the lexical chain construction algorithm and the DBSCAN clustering algorithm are used to extract and cluster the topic words respectively.After that,the business reputation dimension is constructed by extracting the cluster labels.Finally,the experimental verification and case analysis of the user review text in the public comment network prove the feasibility of the method.(2)The comprehensive evaluation of business reputation.Firstly,based on the digging business reputation dimension,a business reputation evaluation index system is constructed,Next the business reputation level is divided and the standard comment set cloud model is constructed as a reference.Then a comprehensive evaluation cloud model of business reputation is built combined with the sentiment analysis.After locating emotional words and emotional polarity words based on HNC theory,the emotional value is calculated by using the duality of the HNC symbol,and the emotional value is input into the inverse cloud generator to generate the digital feature of the cloud model,the comprehensive evaluation cloud model is obtained by using the correlation formula.Finally,by calculating the similarity between the comprehensive evaluation cloud model of the business reputation and the standard comment set cloud model of each level,the evaluation result of the business reputation rating is obtained,and the evaluation result is visually displayed in the form of a simulated cloud map.For the business reputation problem existing in the current social commerce environment,this paper gives a series of methods for the construction of business reputation dimension and comprehensive evaluation of business reputation,which helps consumers make reasonable purchasing decisions according to the business reputation level.At the same time,it provides a reference standard for the business platform to improve its own reputation evaluation system.