考虑时间效应的矩阵分解技术在推荐系统中的应用

Matrix Factorization Techniques with Temporal Dynamics for Recommender Systems

现实生活中,人们的喜好、观念往往随着时间的推移发生变化。人们对事物的看法,在不同的时间、场合也往往不同。即使是同一部电影,心情好的时候与心情坏的时候,给出的评价都不尽相同。因此,在推荐系统设计过程中,如何建立模型捕获这些动态的时间效应,给出更好的推荐结果,显得非常必要。

In real life,people’s preference,ideas tend to change over time.People’s view of things is often not the same at different times and occasions.Even for the same movie,This paper can give dissimilar evaluation in different mood.Therefore,when This paper design the recommend system,it is necessary to consider how to establish a model with temporal dynamics to capture these dynamic effects and give a better recommendation results.