Abstract: To describe the preferences of individual users and to optimize the effect of service recommendation,a QoS Preference-aware Recommendation(QPrefR)scheme based on service reputation was proposed.By taking the multidimensional reputation indexes of services as the analysis object,various forms of reputation were contained which included numerical rating,labeling and evaluation of natural language.Based on fuzzy multiple attribute decision making theory,automatic extraction of user preferences and recommendation of service were achieved.Theoretical analysis and numerical simulation based on real service evaluation data indicated that the proposed model could present the customer's preference requirements effectively.

Key words: reputation, preference, utility prediction, customization

摘要： 为了描述用户的个性化偏好需求、优化服务推荐效果,提出基于服务信誉评价的用户偏好分析与服务推荐模型。该模型以服务的多维信誉指标为分析对象,能够兼容服务信誉评价的各种形式,包括数字评分、标签和自然语言评价,并依据模糊多属性决策理论进行自动的用户服务质量偏好的提取及服务推荐。通过对比分析实际服务数据的实验结果,验证了模型的合理性与有效性。

关键词: 信誉评价, 用户偏好, 效用预测, 个性化