融合时间因素的隐语义模型推荐算法

Recommendation algorithm based on latent factor model of fusion time factor

针对传统推荐算法中存在数据稀疏和精确度不高的问题,提出一种融合时间因素的隐语义模型推荐算法,在隐语义模型中引入时间偏置项体现时间推移对用户兴趣偏好的影响,解决数据稀疏问题的同时降低时间推移造成的误差,结合基于邻域的协同过滤模型求出目标用户推荐列表。采用Movielens1M数据集验证算法的有效性,实验证明该算法与基于用户的协同过滤算法以及基于隐语义模型的推荐算法,能有效解决数据稀疏问题,在准确率、召回率和综合F值上分别比基于用户的协同过滤算法提高1.66%、2.12%、2.04%,比基于隐语义模型的推荐算法分别提高1.38%、1.48%、1.49%,能够进一步提高推荐系统的准确性及推荐质量。

Aiming at the problems of data sparsity and low accuracy in traditional recommendation algorithms,this paper proposes a recommendation algorithm of latent factor model that integrates time factors.The latent factor model introduces a time offset term to reflect the impact of time on user interest preferences,solves the problem of data sparsity while reducing the error caused by time lapse,and combines the neighborhood based collaborative filtering model to obtain the target user recommendation list.The validity of the algorithm is verified by using Movieens1M data set.The experiment shows that the algorithm,together with user-based collaborative filtering algorithm and recommendation algorithm based on latent factor model,can effectively solve the problem of data sparsity.Compared with user-based collaborative filtering algorithm,the accuracy,recall rate and comprehensive F value are 1.66%,2.12%and 2.04% higher,and 1.38%,1.48% and 1.49% higher than the recommendation algorithm based on latent factor model,it can further improve the accuracy and quality of the recommendation system.