Research On The Impact Factors Of The Helpfulness Of The Search Product Online Reviews

With the rapid development of e-commerce,online shopping has gradually become normal.Due to the asymmetric information of online shopping,online commodity reviews become an important source of information for people to obtain goods and services.Meanwhile,it has provided a platform for consumers to share the use of goods experience.Because the characteristics of online search product reviews exhibit massiveness,complexity and variety,mining effective review information requires timeliness and accuracy on the objective.Therefore,to explore the impact factors of the helpfulness of online reviews has important guiding significance for establishing an effective evaluation system.In this paper,the online reviews of six kinds of the search products of totaling one thousand and forty-two were captured on Amazon.com as the research object.The paper was based on the theory of information adoption model,the characteristics of useful information and the theory of consumer purchase behavior.From the source and content of reviews,on the basis of previous studies,we constructed a model of factors influencing the usefulness of reviews.Then we take the review comments length,completeness,timeliness,comments on timeliness,emotional intensity,the source of reliability as independent variables.The commentary content integrity was divided into service information,logistics information and commodity information,with the dependent variable reviews helpfulness.We establish Tobit regression model to explore the impact of reviews helpfulness factors,and to verify it through the correlation analysis of factors.Finally,the evaluation index system was established,and the classification model was established by using the naive Bayes algorithm,the support vector machine algorithm and the C4.5 decision tree algorithm.The comments were divided into two categories: "useful" and "useless".According to the accuracy and efficiency of classification,the optimal classification model was selected.The results show that the more words of comments,the more the reviews mentioned service information and the properties of the products,the better the reviews express the subjectivity of the reviewers,the lower star level of the reviews,the higher perceived usefulness of consumer reviews.And whether the comments mentioned in the logistics information and comments published days on the usefulness of the commentary is not significant.The length of the comment is positively related to the completeness of the content of the comment and the subjectivity of the comment.It has nothing to do with the commentary emotion intensity.The commentator's rank is negatively correlated with the length of the comment and the commodity information in the comment.Support vector machine algorithm classification accuracy is the highest accuracy rate reached 74.28%,F1 value is 0.736,the classification process takes 0.28 seconds;therefore,choose the support vector machine algorithm as the classification model of merchandise online reviews useful.