一种基于时间加权和用户特征的协同过滤算法

协同过滤算法是目前个性化推荐系统中应用最成功的推荐算法之一,但传统协同过滤算法很少考虑到用户兴趣随着时间变化以及用户特征与兴趣的关联性两方面的问题。针对该问题提出了一种基于时间加权和用户特征的协同过滤算法,首先通过定义时间指数函数反映兴趣随时间增长的变化,然后建立用户的特征矩阵,最后采用一种新的相似度度量方法计算出目标用户的最近邻居集合。实验结果表明该算法推荐平均绝对误差(MAE)比传统算法降低了12%,推荐质量较传统算法有明显提高。

Collaborative Filtering Algorithm Based on Time Weight and User's Feature

Collaborative filtering algorithm is one of the most important technologies applied in recommendation system.But traditional collaborative filtering algorithm does not consider the problem of user’s interest drifting over time as well as relevance of user characteristics and interests based on time-weighted and user’s feature of collaborative filtering algorithm is presented in the paper.Firstly definition of exponential function about time reflects changes in user interest over time,and secondly analyzing the matrix of user’s feature,and finally using a new similarity measure to calculate the nearest neighbor set of target users.The experimental results show that the algorithm reduces the mean absolute error(MAE) of recommendation 12% than the traditional algorithm and the quality of recommendation is improved significantly than traditional algorithms.