| With the booming of Internet economy and the gradual deepening of the interaction between the internet and web users in web2.0 ear,the scale of the online reviews in e-commerce platform has growth rapidly.And online reviews has become the most important factors for the decision making of the online shopping users.Huge number of online reviews has put user in front of serious information overload.However due to the lack of proper guidance and specification on users comment behavior,as well as the online review itself is anonymity and its structure and content is not standard and so on,a lot of meaningless reviews flooded.Which creates a big obstacle to online consumers who want to take full advantage of online reviews.How to find the most "helpful" ones out of the huge size of online reviews is really necessary,and is important for improve the online shopping environment.This has received attention from many scholars in related fields and much research has been carried out on the helpfulness of online reviews.Based on these issues,this paper aim at the automatic identification of the usefulness of online reviews on e-commerce platform,by means of machine learning and text classification technology.Through the research of the content of online reviews,we summarized the characteristics of online reviews on e-commerce platform.And analyzed factors that affects the usefulness of online reviews by theoretical analysis.Then we give a definition on the helpful of online reviews,and worked out the measurement of online reviews helpfulness.This creates foundation to support the following feature extraction researches and so on.We achieved the automatic identification of the usefulness of online reviews through SVM binary classification.In the feature extraction process we make use of SVD,LDA and construct dictionary according to our research needs.And it has been proved by experiments that our research method can achieve a good recognition results with feasibility. |
PDF Full Text Request