Collaborative Filtering Algorithm Based on Dynamic Trust Attenuation
Ren Jing-xia, Zhifeng Wu · 2020
The traditional recommendation algorithm has the problems of cold start and poor interpretability, and different neighbors in the collaborative filtering algorithm have different degrees of influence. It also ignores that user interest will drift with time. To solve the above problems, we propose a collaborative filtering algorithm based on dynamic trust attenuation (DTA-CF). Based on the traditional collaborative filtering recommendation algorithm, it examines the common score and time factor to adjust the neighbor selection mechanism and introduces the concept of trust attenuation to redefine the effect of neighbors. Experiments on the real MovieLens movie data set show that the algorithm is reasonable and effective, the accuracy and rationality have been greatly improved, and the recommendation error has been reduced.