基于满意度及特征近似的协同数据融合推荐

Satisfaction and Approximate Feature Based Cooperative Data Fusion and Recommendation

为改善数据融合算法推荐执行效率及推荐结果可靠性,以便为使用者推荐更符合个性化需求的服务,设计一种基于用户满意度及特征近似的协同数据融合推荐算法(SFSTOPSIS)。首先,针对传统相似度定义分辨率不足的问题,基于用户评价置信度、兴趣偏好及特征相似度评价进行改进,并结合用户使用属性对使用者间存在的相似度替代,使之更符合用户真实感受;其次,采用时变权重方式对标准TOPSIS融合进行完善,提高TOPSIS决策融合的时变属性,实现用户相似度数据的有效属性融合;最后,基于标准测试事例进行实验对比,显示所提SFSTOPSIS算法可有效提高服务推荐精度,具有一定应用价值。

In order to improve the efficiency of the data fusion algorithm and the reliability of recommendation results, so as to recommend more personalized services to meet the needs of users, the satisfaction and approximate feature based cooperative data fusion and recommendation algorithm is proposed here. Firstly, aiming at the problem of low resolution of traditional similarity definition, here the improvement is done for it based on the user＇s evaluation confidence, preferences and feature similarity evaluation, and combined with the user using attribute, the similarity between the users is replaced, which makes it more in line with the user＇s real feeling; Secondly, the time-varying weight is used to improve the standard TOPSIS fusion algorithm, and also improve the time-varying properties of TOPSIS decision fusion, which realizes the integration of effective properties of user similarity data; Finally, based on the standard test case, the results show that the proposed SFSTOPSIS algorithm can effectively improve the service recommendation accuracy, and has a certain application value.