Research On The Approach Of Automatically Identifying Usefulness Of Search Product Reviews Based On SVM

Development of Internet technology makes online shopping become a new form ofconsumption, online shopping has over80million users. Because of the rapid spread ofthe Internet information, online shopping users can see a lot of information online productreviews, online product reviews can help consumers make effective purchasing decisions,but the online product reviews quality varies greatly, so that online shopping can not beuseful to quickly identify their useful product reviews. This paper studies the usefulnessof product reviews, product reviews usefulness in analyzing factors based on the use oftext classification techniques to achieve product reviews usefulness of automaticidentification.Firstly, through summarizing product reviews usefulness as well as text classificationtechnology research, we draw the relationship between comment text effectivenessclassification and text categorization relationship, thus gives solutions of automaticidentification method of product reviews usefulness. Then by analyzing the characteristicsof online reputation and relationship of online reputation and online product reviews,summarize the characteristics of online product reviews, and give the definition ofproduct reviews usefulness and criteria for defining useful product reviews. Secondly,from product reviews sources and comment text two angles, analysis factors of productreviews usefulness, determine the evaluation of product review usefulness and everyspecific indicators, this establish the foundation of constructing product review textfeature vectors. Finally, achieve automatic identification of product reviews usefulness.Identifying the usefulness of product reviews automatically means classifying the productreview into "useful" and "useless", building product reviews text feature vectors, usingsupport vector machine algorithm to achieve comment text classification, automaticallyidentifying the usefulness of product reviews.This research has important theoretical and practical significance on online shoppingusers, e-commerce sites and enterprise, and will promote the development of e-commercewebsite commenting system, make the commenting system more perfect, and providebetter service for online shopping users.