Model Research Of Infuliencing Factors On Online Commodity’s Attrction And Helpfulness

The problem of information overload occurs when consumers read reviews. To solve this, filtering and sorting for reviews is indispensable. Consumers’ reading process can be divides into notice stage and comprehend stage. Correspondingly, as dependent variables, review attraction and review helpfulness are put forward. Exploring the influencing factors of these dependent variables can provide ways for filtering reviews.Considering the characteristics of typical online review systems, information can be parted as explicit information and implicit information. Review attraction is assumed to be mainly influenced by the former and review helpfulness is influenced by both. Depend on literature analysis, theoretical hypothesis and conceptual model are raised. I collected review samples for empirical study, including 4 categories and 8 commodities. Text analysis is used to construct variables and regression model is used to test conceptual model for influencing factors. Moreover, this study explores the applications of analysis results. Main research results of this study are shown as follows.1. Review extremity, review reliability and reviewer rank significantly influence review attraction. Reviews with negative extremity, numerous words and high reviewer rank are easily to gain more attraction.2. Review extremity, reviewer rank, mixed subjective property and mixed sentiment significantly influence review helpfulness. Reviews with negative extremity, high reviewer rank, mixed subjective property and mixed sentiment probably have high helpfulness. Besides, for search commodities, review extremity and mixed sentiment’s influence on review helpfulness is more significant than experience commodities.3. For application, this study improved two strategies for filtering commodities. Based on 4 significant factors, Support Vector Machine and Random Forest are attempted to predict review helpfulness and the predict accuracy is bigger than 75%. Meanwhile, several practical suggestions are proposed in the end.