Research On Commodity Evaluation Model Based On The Helpfulness Of Online Reviews

With the rapid spread of online shopping,more and more users choose to purchase goods and services online to meet their individual needs.What has changed is that the number of online reviews has increased dramatically.Studies have shown that online reviews can influence a user's purchasing decision.Therefore,it is very important and necessary to help users to extract product-related information from massive commentary information to help users in making decisions.After reviewing the literature on commodities evaluation,the helpfulness of online reviews,commodity feature extraction,sentiment analysis,and other topics,the comprehensiveness and relevance of online reviews,sentiment intensity,review text length,user scores,and time span of reviews are selected,and the indicators are quantified using commodities feature extraction and sentiment analysis.Then,combined with the analytic hierarchy process and TOPSIS ranking algorithm,the quantitative value of the helpfulness of online reviews is calculated.In the evaluation of a commodity,the sentiment value of each review in each characteristic dimension of the commodity is calculated,and the sentiment value is corrected by using the helpfulness quantitative value of the review to distinguish its reference value for the process of commodity evaluation.Finally,based on the entire data set,this paper establishes a fuzzy evaluation matrix for each commodity id and calculates the commodity evaluation score.By comparing the relative sizes of the commodity evaluation values,a reference angle can be provided for the user's decision-making,and then the evaluation of the commodity is completed on the basis of fully considering the user experience and the user's voice.There are three innovations in this paper:(1)In the past research on commodity evaluation,although the existing research is gradually improving,the evaluation did not deal with the data itself,and there were comments that took a long time to participate in the evaluation process.For this reason,in constructing the fuzzy evaluation matrix,this paper uses the helpfulness of the review to modify the matrix,which is more reasonable;(2)The indicators of the number of commodity features and the number of sentiment words in the helpfulness index of online reviews are optimized.Different from the index of the number of commodity features and the number of sentiment words used to calculate the helpfulness of online reviews in the past,this paper improves them into a method that comprehensively considers the number of commodity features and the importance of commodity features,and calculates the intensity of sentiment using sentiment analysis.This improvement not only refines the impact of its indicators,but also grounds sentiment measurement.This lays the foundation for the improvement and further exploration of the indicators;(3)This paper uses the sentiment dictionary and the method of defining rules to calculate the sentiment intensity value,and uses the F1 value as an index for the sentiment classification result,by using a random forest combined with different vectors.The sentiment classification,the F1 value of the result is compared with the sentiment analysis method in this paper,which proves the accuracy and reliability of the method in this paper.