基于用户兴趣变化的协同过滤推荐算法

协同过滤算法是在众多应用领域中最成功的个性化推荐技术之一,但传统协同过滤算法不能及时反映用户的兴趣变化,致使预测结果不准确。针对这个不足,提出一种基于用户兴趣变化的改进协同过滤算法。改进算法提出一种基于时间的权重函数,用于研究用户在不同时间段的兴趣变化,通过用户兴趣之间的相似性,最后生成推荐结果。实验结果验证了改进算法在推荐的准确性方面得到显著提高。

Study of collaborative filtering recommendation algorithm based on users＇ interest change

Collaborative filtering algorithm is one of the most successful technologies for providing personalized rec- ommendations in various application areas. However,the traditional collaborative filtering algorithms do not reflect the change of users＇ interest in time. For this reason,the prediction outcome of the traditional collaborative filte- ring approach may be inaccurate when users＇ interest has changed. Considering this defect,an improved collabora- tive filtering algorithm is proposed to take users＇ interest change into account in the paper. A time-based weighting function is proposed to study the changes of users＇ interest in different time periods in the improved algorithm and results in recommendation by comparing the similarity between users＇ interests. Experiments results show that the new algorithm improves significantly in recommendation effectiveness.