Bi-Group Bayesian Personalized Ranking from Implicit Feedback
Haifeng Zhao, Xingjun Wang · 2019
The personalized ranking is to provide a user with a ranked list of items, which is also called item recommendation. Item recommendation from implicit feedback has attracted much attention in recent years due to its broad application in real-world scenarios. In this work, we introduce the concept of Bi-Group: support group and opposition group. The members in the support group will support the user's positive feedback by rewarding item preference, and the members in the opposition group will object to the user's positive feedback by penalizing item preference. Based on Bi-Group, we present Bi-Group Bayesian Personalized Ranking(BiGBPR) algorithm that exploits these two groups of users to reward and penalize user's actions. Experiments on three real-world datasets show that BiGBPR can significantly improve the performance of item recommendation.