QINR: A Quantum-Inspired Network for Interpretable Review-based Recommendation
Tingsan Pan, Yuexian Hou, Tian Tian, Zan Li · 2021 2nd International Conference on Electronics, Communications and Information Technology (CECIT) · 2021
Recent studies in recommender systems have managed to achieve significantly improved performance by leveraging reviews for rating prediction. However, convolutional neural networks (CNNs) have been widely adopted in recent methods, which are difficult to intuitively understand and interpret. To make the recommendation process more transparent and interpretable, we seek to model reviews by the mathematical framework of quantum physics rather than general deep learning methods. In this paper, we propose a Quantum-Inspired Network for interpretable review-based Recommendation (QINR), which models reviews with the well-designed mathematical formulations in quantum physics and further predicts user ratings for items. Specifically, reviews are represented by density matrices and a set of measurement operators are applied to the density matrices. Experiments on the Amazon datasets demonstrate that QINR outperforms recent recommendation methods in terms of mean square errors (MSE).