Trust-aware Collaborative Denoising Auto-Encoder for Top-N Recommendation

Both feedback of ratings and trust relationships can be used to reveal users' tastes for improving recommendation performance, especially for cold users. However, both of them are facing data sparsity problem, which may severely degrade recommendation performance. In this paper, we propose to utilize the idea of Denoising Auto-Encoders (DAE) to tackle this problem. Specially, we propose a novel deep learning model, the \textit{Trust-aware Collaborative Denoising Auto-Encoder} (TDAE), to learn compact and effective representations from both rating and trust data for top-N recommendation. In particular, we present a novel neutral network with a weighted hidden layer to balance the importance of these representations. Moreover, we propose a novel correlative regularization to bridge relations between user preferences in different perspectives. We also conduct comprehensive experiments on two public datasets to compare with several state-of-the-art approaches. The results demonstrate that the proposed method significantly outperforms other comparisons for top-N recommendation task.