Reviewer Reputation Ranking In Product Review Systems

Abstract

With the rapid growth of the Internet and the Web technology, individuals, public and private organizations are concerned in marketing their products and services on the Web. In addition to the advertising on the Web, vendors have given the opportunity to the consumers to evaluate and comment on products and services in review systems. Most of the review systems pay less attention on verifying the identity of reviewers and the content of reviews. Reviewers can freely post reviews without going through a rigorous editorial process for factual veri cation in most of these review system. This leads to the presence of uninformative, biased or even false information, which make it time consuming and challenging for the consumers to nd accurate information. If reviewer reputation can be measured, the consumers can form their own opinion about a particular product much faster and easier, depending on the reviews from reputed reviewers. Most of the current works assess the reviewer credibility based on the help- fulness voting. This thesis presents the proposed reputation model devised for quantitatively measure reviewer reputation based on the source credibil- ity theory, without applying direct judgments by reviewers. Based on the identi ed key factors namely expertise, low review content similarity, low rat- ings deviation, single product group multiple high ratings, and single product group multiple low ratings, a multiple linear regression was applied to predict reviewer reputation. Our experimental results show that the proposed rep- utation model outperforms existing methods in measuring the reputation of reviewers to help search for credible reviews. The scheme is simple, e ective, and applicable to most of the online review systems.