基于用户兴趣迁移的隐语义模型推荐算法

A Recommendation Algorithm of Latent Factor Model Based on User Interest Migration

协同过滤算法一般根据用户的评价信息来推测用户的喜好,但受到数据稀疏问题的影响,许多时候无法得到较为理想的推荐结果。除此之外,一般协同推荐算法忽略了用户兴趣的动态变化。本文所提出的基于改进的隐语义模型(TLFM算法)则将用户的兴趣喜好的改变考虑进来。具体的,在传统的隐语义模型的基础上加入对时间信息的权衡,选取适当的权重,有效地模拟用户兴趣随时间的变化状况,根据用户兴趣变化来决定推荐值,可以很好的提高推荐准确度。

The collaborative filtering algorithm generally estimates the user’s preference based on the user’s evaluation information, but is affected by the data sparse problem, and many times it is impossible to obtain a better recommendation result. In addition, the general collaborative recommendation algorithm ignores the dynamic changes of user interest. The improved implicit semantic model(TLFM algorithm) proposed in this paper takes into account the changes in user interest preferences. Specifically, on the basis of the traditional implicit semantic model, we add the trade-off of time information, select appropriate weights, effectively simulate the change of user interest over time, and determine the recommended value according to the change of user interest, which can improve the recommendation accuracy.