Abstract: User rating on products or information services includes reviews and scores, and reflects user behavior      information, such as interest, opinions and preferences. In order to represent the degrees of user preferences on products inherently and quantitatively, starting from the massive rating data, this paper defines user preference based on the idea of marginal utility. Then, this paper describes the uncertainties of relevant influence factors on user preferences and the mutual relationships among these factors based on the D-S evidence theory. Taking the vocabulary in a review, the vocabulary including positive/negative words and the score as the evidence of user preference respectively, this paper gives the operator for combining the relevant factors jointly, as well as the computation method and mechanism for discovering user preferences based on MapReduce. The experimental results on correctness, execution time, speedup and parallel efficiency verify the effectiveness of the method proposed in this paper.

Key words: massive rating data, user preference, D-S evidence theory, evidence fusion, MapReduce

摘要： 用户对商品和信息服务的评价包含评论和评分，富含了用户的兴趣、观点和偏好等行为信息。以真实和量化地反映用户对商品的喜好程度为目标，从海量的用户评价数据出发，基于边际效用定义用户偏好，基于D-S证据理论描述影响用户偏好的各影响因素的不确定性以及各因素之间的相互关系；以评论中的各词汇、包含正面/负面词汇的评论和评分作为用户对商品偏好的“证据”，给出了综合考虑各影响因素的联合算子，以及基于MapReduce的计算方法和用户偏好发现机制。针对正确性、执行时间、加速比和并行效率等指标进行实验，结果验证了所提出方法的有效性。

关键词: 海量评价数据, 用户偏好, D-S证据理论, 证据融合, MapReduce