一种结合评分时间特性的协同过滤推荐算法

摘要

用户评分是协同推荐算法实现未知评分预测的主要依据，传统协同推荐算法一般只利用评分的数值，而忽视评分产生时间对推荐的作用，但是评分时间特性对推荐系统准确性的影响不容小觑。本文针对这个问题，以传统协同过滤推荐算法为基础，从评分时间角度对推荐算法的相似度计算和评分预测过程进行改进，提出了一种结合评分时间特性的协同过滤推荐算法。算法依据用户对项目的评分及时间计算出一个时间因子，并将时间因子融入到相似度的计算中，使推荐给目标用户的项目更加合理。通过实验进行该算法与现有协同推荐算法的对比，验证了该算法在提高推荐准确性方面的有效性。%User rating is the main basis for cooperative recommendation algorithm to achieve the unknown score prediction. Traditional collaborative recommendation algorithm usually only use the score value while ignoring the ef-fect of score generation time on recommendation. However the impact of scoring time characteristics on the accuracy of recommendation system can not be underestimated. In this paper, we propose a new collaborative filtering recommen-dation algorithm based on the traditional collaborative filtering recommendation algorithm, which improves the similar-ity calculation and the score forecasting process of the recommendation algorithm from the scoring time angle. The algorithm calculates a time factor based on the user’s score and time of the project, and integrates the time factor into the calculation of the similarity, which makes the recommendation to the target user more reasonable. The experiment is compared with the existing collaborative recommendation algorithm to verify the effectiveness of the algorithm in im-proving the recommendation accuracy.