Abstract: Traditional memory-based collaborative filtering algorithm has the problem of bad scalability，while the model-based collaborative filtering algorithm，due to lagged updating hysterics，has the problem of bad recommendation. To solve the above problems，a collaborative filtering algorithm based on real-time users feedback was proposed，which achieved that recommender system can finish the real-time updating of the model data when a new rating was submitted by active user. Hence, recommender system can reflect the changing of user interest accurately. The experimental results indicate that the algorithm can improve the recommendation accuracy efficiently and reduce the recommendation time significantly.

Key words: collaborative filtering, similarity feedback mechanism, mean absolute error, mean access time, mean recommended time

摘要： 传统的基于内存的协同过滤算法存在可扩展性不足的问题，而基于模型的协同过滤算法由于模型数据的滞后，造成推荐质量不高。针对以上情况，提出一种基于用户实时反馈的协同过滤算法，该算法在用户提交项目评分之后能实现对推荐模型数据的实时更新，从而更精确地反映用户的兴趣变化。实验结果表明，该算法能够有效地提高推荐精确度并且大幅地缩短了推荐时间。

关键词: 协同过滤, 相似性反馈机制, 平均绝对误差, 平均评分时间, 平均推荐时间