An Approach to Rating Prediction for Personality Recommendation via Attention Mechanism and Denoising Autoencoder
Hui Zhu, Zhongshen Qian, Zulai Ye, Ding Zhang · 2022
Rich personalized information, trust and distruct, which is fully utilized can improve the quality of rating prediction for personality recommendation. In this work, we propose a novel model of Attention Mechanism and Denoising Autoencoder Recommender, namely AMDAERec, to explore the nonlinear features between users who have diversity relationships and items which are interacted with these users. We also compare this model with eight state-of-the-art personality recommendation algorithms based on the Epinions dataset. Experimental results show that our approach can make full use of users' trust or distrust relationship information and capture the nonlinear features between users and items. Finally, we effectively improve the accuracy of rating prediction.