使用贝叶斯网络的社交网络信任驱动推荐方法

发布时间：2018-08-07 22:05

【摘要】：为处理推荐行为来源复杂、路径多样、不信任陌生推荐等问题,提出一种在社交网络中信任驱动推荐方法。该方法利用贝叶斯网络,计算用户评分的先验概率分布以及朋友之间的联合条件概率,预测用户在该环境下的评分并将推荐给用户。在信任驱动推荐过程中,预测评分既考虑到用户的偏好,也考虑到用户的社会关系;此外,用户的信息交换只限于朋友之间,能够有效保护用户的隐私。实验结果表明,所提出的推荐方法在预测准确率和推荐覆盖率上具有良好的性能。
[Abstract]:In order to deal with the problems of complex sources of recommendation behavior, multiple paths and distrust of unfamiliar recommendation, a new method of trust-driven recommendation in social networks is proposed. The method uses Bayesian network to calculate the prior probability distribution of the user score and the joint conditional probability among friends, and predicts the user's score in this environment and recommends it to the user. In the process of trust-driven recommendation, the prediction score takes into account the user's preference as well as the user's social relationship. In addition, the user's information exchange is limited to the friends, which can effectively protect the user's privacy. The experimental results show that the proposed recommendation method has good performance in prediction accuracy and recommendation coverage.
【作者单位】：南京邮电大学计算机学院;南通科技职业学院信息工程系;南京邮电大学通信与信息工程学院;
【基金】：国家自然科学基金(No.61501253,No.60905040) 中国博士后科学基金项目(No.2013M531393) 江苏省“六大人才高峰”第十一批高层次人才选拔培养资助项目(No.XXRJ-009) 江苏省基础研究计划(自然科学基金)项目(No.BK20131382,No.BK20151506) 江苏省博士后科研资助计划项目(No.1102102C) 江苏省2014年度普通高校研究生实践创新计划项目(No.SJZZ_0107)
【分类号】：TP391.3

【相似文献】