基于用户特征和时间权重的协同过滤算法

为了解决传统算法中寻找最近邻居不准确和用户兴趣随时间变化而迁移的问题,提出一种基于用户特征和时间权重的协同过滤算法。文中首先把Movie Lens数据集中用户特征信息数字化,求出用户特征相似性,将其加入到修正的余弦公式中,得到一种新的用户相似度,以找到更加准确的最近邻居集;然后通过引入时间函数来反应用户的兴趣迁移,再根据预测评分公式来获得更加准确的预测评分;最后给用户生成一个较可靠的推荐结果。实验结果表明,该方法取得了较好的效果且平均绝对误差(MAE)值达到72.57%。

Collaborative Filtering Algorithm Based on User Feature and Time Weights

In order to solve the problem of user interest migration and inaccurate recommendation of traditional cooperative filtering algorithm, this paper proposes a collaborative filtering algorithm based on user characteristics and time weight. The algorithm first digitizes the user characteristic information of the MovieLens data set and obtains the user characteristic similarity. Then it adds to the modified cosine formula to get a new user similarity, finding a more accurate nearest neighbor set. Followed by the introduction of a time function to reflect the user's interest migration, which gives the only non-linearly decreasing weight of each score to obtain a more accurate predictive score. And finally produce a more reliable recommendation list. The experimental results show that the method has achieved good results with an average absolute error (MAE) of 72. 57％.