Research On The Influence Mechanism Of Online Review Helpfulness Based On Text Mining

With the development of e-commerce,more and more consumers choose to buy goods on the Internet.Unlike traditional physical consumption,consumers cannot obtain intuitive feelings and trial experiences when shopping online.As a result,product reviews,as one of the most trusted sources of information for consumers,are becoming increasingly important in their daily decisions.Unlike product information provided by merchants,consumer reviews can provide consumers with first-hand experience information about products.At the same time,since the source of information is consumers who have previously purchased the product,online reviews also are considered more reliable by potential consumers.Therefore,the information provided in online reviews has the potential to reduce the risks and uncertainties inherent in consumers' online purchase decisions.However,online reviews have grown exponentially,causing information overload for consumers,making it difficult for consumers to obtain the information they really need.How to find useful reviews has become a common concern for online retailers and academics.Although there are a large number of literatures that analyze the influencing factors of the helpfulness of online reviews,the existing literature still has the following deficiencies regarding the textual characteristics of reviews:(1)The mechanism of the numerical and textual characteristics of online reviews affecting review helpfulness is not clear.At present,only the objective features of the review text are concerned,and the fact that reviewers express their personal subjective view in the review is ignored.(2)The research on the impact of the text similarity between review title and review content on the helpfulness of the review is insufficient and lacks theoretical support.In view of the above deficiencies,this study focuses on the following two questions:(1)Are there any differences between the numerical features and text features of online reviews in affecting review helpfulness?(2)Does the text similarity of review title and review content affect review helpfulness,and will the title sentiment and content sentiment strengthen or weaken this relationship? Focusing on the above two issues,this paper studies how online review text affects consumers' perceived review helpfulness by combining text mining technology and empirical analysis methods.First,this study analyzes whether the numerical and textual features of online reviews affect customer perception of review usefulness and the moderating role of review types.Through the analysis of 30,338 product reviews from Amazon.com,this study found that(1)the numerical and text features of online reviews have different effects on review helpfulness,and this difference is moderated by the type of reviews.(2)When predicting the usefulness of online reviews,the numerical features of reviews play a better role than textual features,and review length,one of the numerical features of a review,has the best effect in predicting the helpfulness of online reviews.Second,this study analyzes the impact of text similarity between review title and review content on review helpfulness,and explores the moderating effects of title sentiment,content sentiment and their consistency on the relationship between text similarity and review helpfulness.By analyzing 127,547 product reviews from Amazon.com,this study found that(1)the more similar the title and content are in the text,the more helpful a review is.(2)The title sentiment negatively moderates the relationship between the title-content text similarity and review helpfulness.(3)When title sentiment is consistent with content sentiment,the positive effect of the text similarity on review helpfulness will be strengthened.There are three main theoretical contributions in this paper.First,the text mining method is applied to the field of online reviews,and the effect of the text characteristics of online reviews on consumers' perception of the value of reviews is analyzed.Specifically,this study uses text mining methods such as text classification,text similarity measurement,and text sentiment analysis to analyze text characteristics in online reviews.Second,the relationship and differences between numerical characteristics and textual characteristics in product reviews were clarified,and a reasonable explanation was given for the divergences regarding the influence of numerical characteristics and textual characteristics on review helpfulness.Third,the impact of the similarity between review title and review content on consumers' perception of the value of reviews is discussed.The moderating effects of title sentiment,content sentiment,and sentiment consistency on the relationship between text similarity and review helpfulness are also considered.