一种结合共同邻居和用户评分信息的相似度算法

Similarity Algorithm Based on User's Common Neighbors and Grade Information

随着互联网的发展,推荐系统逐步得到广泛应用,协同过滤(CF)是其中运用得最早、最成功的技术之一。CF首先根据用户间的相似度,找出每个用户的近邻;然后根据目标用户近邻的评分预测目标用户的评分;最后把预测评分较高的项目推荐给目标用户。因此相似度计算方法直接关系到预测结果的准确性,对推荐起着至关重要的作用。目前,学者们已从不同的角度提出了各种各样的相似度计算方法,其中共同邻居算法(common-neighbors)是一种简单有效的方法。但此法仅考虑了两用户间的共同邻居数,忽略了用户的具体评分信息。针对这个问题对共同邻居算法进行了改进,同时考虑了共同邻居数和用户的评分信息。实验结果表明,改进的共同邻居算法在一定程度上可提高评分预测的准确性。

With the rapid growth of Internet,recommender systems have been used in many fields,and collaborative filtering（CF） is one of the earliest and the most successful ones.CF method usually identifies the neighborhood of each user based on similarity between two users; then predicts items＇ rating by integrating ratings of target user＇s neighbors,and lastly those items with higher predicted score are recommended to target user.So similarity plays an important role and affects the accuracy of the prediction.Up to now,various similarity measures have been proposed by resear-chers from different aspect.And common-neighbor algorithm is a simple and efficient method.However,common-neighbor algorithm just considers the number of common objects scored by two users,doesn＇t consider the user＇s grade information.In this paper,an improved algorithm based on common-neighbor and user＇s grade information was proposed.Experimental results indicate that improved common-neighbor algorithm can obtain rather good predicting results.