Research On The Helpfulness Estimation Of Online Reviews: Considering The Reviewers’ And Voters’ Factors

With the fast development of web2.0applications and e-commerce, the role ofonline reviews gets more and more attention. Online reviews are generated byordinary users and provide more objective information about related products andservice; Consumers can use these reviews to assist their purchase decision, and sellerscan collect customer feedback from these reviews. As the great value contained in theonline reviews, the need of analyzing these reviews seems urgent.However, analyzing online reviews faces many challenges. Frist, the volume ofonline reviews is huge, and there are hundreds of reviews for only one product. As thegreat number of online reviews, we can’t analyze them manually and automatetechnoly becomes necessary. Additionally, online reviews are written by ordinaryusers and the quality can’t be controlled, so the helpfulness varies a lot. Therefore,many researchers focus on the problem of helpfulness estimation of online reviews,trying to find small set of highly helpful reviews.As to the research of helpfulness of online reviews, the main objective is to findall kinds of factors that affect the helpfulness and adopt sophisticated method tovalidate. At present, scholars have done much work on the factors of review text andrelated product, and they start to focus on the effect of reviewer related factors. Then,we treat one former study as the baseline, and utilize more factors of the reviewer toimprove the helpfulness estimation.Besides, most former studies use the aggregated results of helpful votes as thecriteria. In our study, we try to directly estimate the helpful votes and firstly considerthe voters’ factors. We utilize three aspects of voter related factors, including historyvotes of the voter, trust relation between the voter and reviewer and trust linksbetween the voters. At last, we compared experiment results of several models talkedin my study.My research provides a fresh new view for the helpfulness estimation of onlinereviews, and the results in this paper can also offer some references for the real onlinereview system.