结合用户判断力和相似性的协同推荐算法

Collaborative Recommendation Algorithm Combining User's Judging Power and Similarity

作为解决信息超载问题的有效方式,协同过滤技术已被成功地应用到推荐系统。为进一步提高协同过滤算法的性能,首先利用用户评分的历史信息,估计用户的判断力。接着结合用户间的相似性,提出一种改进的协同推荐算法。实验结果表明用户的判断力可与用户的推荐能力正相关,也验证了用户判断力深入抽取用户评分信息以及影响用户采纳某项推荐结果的因素,以更好地刻画用户之间的相似性,从而提高算法的推荐准确度。

As an effective way to solve information overload,collaborative filtering（CF）technology has been successfully used in recommendation system.To improve the performance of CF algorithm,first,this paper evaluated user＇s judging power based on historical scoring.Then combining user＇s judging power and similarity,an improved collaborative recommendation algorithm was proposed.Experimental results show that judging power has positive correlation with recommendation abilities of users,which also verify that judging power extracts the depth information from historical scoring and factors influencing a user on adopting recommendation results.So it can characterize the similarity between users better and improve the accuracy of a recommendation algorithm.