中文摘要：

互联网中对产品和服务的评价越来越受到重视，因为评价能够消除消费者的不确定性，辅助其做出购买决策。大多数在线购物网站中用户的评价包括评级和评论。现有的评价反馈系统和评价研究往往只单独关注评价者之间的评级或评论，而忽略了两者之间的有机统一。评价者的评级并不一定反映评价者的真实评价，很多评价者更倾向于选择评论文本来表达自己的真实情感。本文以从淘宝网抓取的852071条评价数据为基础，通过分析评价者评级和评论之间的不一致性，结合RFM模型，考虑评级和评论两种信息形成的评价效用，提出了RFMA模型来计算评价者的总体评价效用。并据此对好的与不好的评价者进行区分，进一步为消费者的购买决策提供支持。通过对评价者的总体分析可以得出，本文提出的机制更加具有可用性和有效性。

英文摘要：

Appraisals for products and services are increasingly important on the Internet, as they eliminate consumers＇uncertainty, and help them to make purchase decision. Raters＇ appraisals for products are di- vided into ratings and comments in most online shopping sites. The existing online reputation system and appraisal studies tend to focus on the user rating or comment respectively, but ignore the organic unifica- tion between them. User ratings do not fully reflect users＇ real evaluation, as they are inclined to express their true feelings by comments. On the basis of the 852071 appraisal captured from Taobao, this paper proposes RFMA model to calculate raters ＇ appraise quality, which combines RFM model and considers two kinds of information containing rating and comment by analyzing the inconsistency of rating and comment. Then the good raters and bad raters are distinguished, and further support for consumer purchase is provided. The proposed RFMA model finds a new mechanism for measuring raters＇ effectiveness. It can be used as a basement for shopping platform to classify the raters, and provide a new way of thinking to further improve the existing online reputation system. Through analyzing all of the raters, it can be concluded that the mechanism of combining the comments is more available and effective.