Abstract: The rewards on favorable comment can be regarded as a transaction for reputation that consumers give favorable comment and sellers give cash back as a reward. While current academic research mainly focuses on the reputation mechanism of online reviews or the identification strategy for the spam reviews, little attention has been paid to the reputation mechanism of the certain comments with the character of bribe generated by the rewards on favorable comment. The data from the survey of online and offline is used to construct a dynamic reputation modified model of online reviews. In order to obtain the change of score difference, Taobao and tmall were followed up twice in 2016 and 2017 respectively. The model is designed to elucidate how the induced positive reviews work on the final reputation evaluation. According to the results, the rewards on favorable comment is not necessarily to improve the store's reputation score, but only to improve the reputation of low reputation stores. Compared with the reputation damage to high reputation stores, this promotion is very limited. Although praise return is common in different commodity categories, there is no positive incentive for consumers' comment intention from small return, and there is no significant statistical relationship between the willingness to participate in praise return and the behavior of choosing praise return. Anonymous environment often makes consumers more likely to make behaviors that they don't agree with but can benefit from. Whether the merchants participate in the high praise return does not depend on their initial reputation. In the data of the high praise return, we not only observed the low rating stores, but also observed the high rating stores. Further contingency analysis also shows that there is no significant correlation between whether the merchants participate in the return of praise and the initial reputation level of the store, which shows that the motivation of the merchants to participate in the return of praise is similar - to improve the reputation score of the store or at least hedge part of the negative comments, rather than belong to the risk aversion behavior of low reputation stores.

Key words: the rewards on favorable comment, spam reviews, reputation mechanism