Neural Attentional Rating Regression with Review-level Explanations

By Chong Chen, Min Zhang, Yiqun Liu, Shaoping Ma. 2018

644 citations recorded by the source corpus.

Reviews information is dominant for users to make online purchasing decisions in e-commerces. However, the usefulness of reviews is varied. We argue that less-useful reviews hurt model's performance, and are also less meaningful for user's reference. While some existing models utilize reviews for improving the performance of recommender systems, few of them consider the usefulness of reviews for recommendation quality. In this paper, we introduce a novel attention mechanism to explore the usefulness of reviews, and propose a Neural Attentional Regression model with Review-level Explanations (NARRE) for recommendation. Specifically, NARRE can not only predict precise ratings, but also learn the usefulness of each review simultaneously. Therefore, the highly-useful reviews are obtained which provide review-level explanations to help users make better and faster decisions. Extensive experiments on benchmark datasets of Amazon and Yelp on different domains show that the proposed NARRE model consistently outperforms the state-of-the-art recommendation approaches, including PMF, NMF, SVD++, HFT, and DeepCoNN in terms of rating prediction, by the proposed attention model that takes review usefulness into consideration. Furthermore, the selected reviews are shown to be effective when taking existing review-usefulness ratings in the system as ground truth. Besides, crowd-sourcing based evaluations reveal that in most cases, NARRE achieves equal or even better performances than system's usefulness rating method in selecting reviews. And it is flexible to offer great help on the dominant cases in real e-commerce scenarios when the ratings on review-usefulness are not available in the system.

Subfields: Artificial Intelligence, Information Systems

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. Xiangnan He
2. Xu Chen
3. Yongfeng Zhang
4. Xiang Wang
5. Xing Xie
6. Tat-seng Chua
7. Meng Wang
8. Fangzhao Wu

In Information Systems

1. Yongfeng Zhang
2. Xiangnan He
3. Xu Chen
4. Xiang Wang
5. Xing Xie
6. Tat-seng Chua
7. Meng Wang
8. Le Wu

Citations in the corpus

• Explainable Recommendation: A Survey and New Perspectives (2018, methodology): “[2017]), attention mechanism (Chen et al. [2018c]), memory networks (Chen et al. [2018b]), and many others, and they have also been applied to different recommendation tasks regarding explainability, such as top-n recommendation and sequential recommendation. Based on LSTM models, the system can eve”
• Personalized Fashion Recommendation with Visual Explanations based on Multimodal Attention Network: Towards Visually Explainable Recommendation (2019, methodology): “In particular, D-Attn [31] and NARRE [5] leveraged attention mechanism to automatically identify important review information under the supervision of user-item ratings.”
• Graph Neural Networks for Social Recommendation (2019, methodology): “To alleviate the limitation of mean-based aggregator, inspired by attention mechanisms [3, 38], an intuitive solution is to tweak αi to be aware of the target user ui , i.”
• A Capsule Network for Recommendation and Explaining What You Like and Dislike (2019, methodology): “mantic information. NARRE argues that reviews should be treated differently upon different user-item pairs and proposes to employ an attention mechanism to select useful reviews for rating prediction [4]. Likewise, MPCN exploits a pointer-based co-attention schema to enable a multi-hierarchical info”
• A Reinforcement Learning Framework for Explainable Recommendation (2018, methodology): “Embedded methods [7]–[10], [12], [13] integrate the explanation generation process into the construction of a recommendation model.”
• Recommender Systems in the Era of Large Language Models (LLMs) (2023, positive): “Meanwhile, a neural attention framework NARRE is introduced to simultaneously predict users’ ratings towards items and generate review-level explanations for the predictions [33].”
• Interpretable deep learning: interpretation, interpretability, trustworthiness, and beyond (2021, background): “Recent works on explainable recommendation systems can be categorized following our proposed taxonomy, and most of them focus on designing interpretable modules [36, 102, 147, 162].”
• Generate Natural Language Explanations for Recommendation (2021, methodology): “Since user textual reviews are informative and be er re ect user preferences, a lot of research explored the possibility of incorporating user reviews to improve the recommendation quality [3, 20, 22, 43, 49] and recommendation explainability [4, 6, 18, 29, 39, 47], which helps to enhance the e ecti”
• A Survey on Accuracy-Oriented Neural Recommendation: From Collaborative Filtering to Information-Rich Recommendation (2021, background): “such as review-level explanations [109], [110].”
• Personalized Reason Generation for Explainable Song Recommendation (2019, background): “In review-based explainable systems, Chen et al. [37] introduce an attention mechanism to explore the usefulness of reviews and propose a neural attentional rating regression model for recommendation.”
• A Context-Aware User-Item Representation Learning for Item Recommendation (2017, methodology): “[6] propose an attention-based deep learning model (NARR) to compute the usefulness of reviews for recommendation.”
• Dynamic Explainable Recommendation Based on Neural Attentive Models (2019, methodology): “• NARRE: This is a state-of-the-art explainable recommendation method (Chen et al. 2018a), which has been verified to outperform many promising algorithms including NMF, SVD++, HFT and DeepCoNN on Amazon and Yelp datasets.”
• Explainable Recommendation through Attentive Multi-View Learning (2019, methodology): “We set k = 32 for EFM, CKE, DeepCoNN, NARRE and DEAML.”
• MMALFM (2018, background): “Researchers also attempt to apply deep textual modeling on reviews for recommendation [9, 10, 12, 64, 65, 67].”
• Incorporating Interpretability into Latent Factor Models via Fast Influence Analysis (2018, background): “Existing works on enhancing recommender systems with interpretability can be classified into different categories based on the type of recommendation models, including latent factor models [1, 3, 23, 26, 39], memory-based models [29, 32], graph-based models [5, 15, 35], deep learning models [8, 24, ”
• The FacT: Taming Latent Factor Models for Explainability with Factorization Trees (2019, methodology): “Another possible way of explanation generation is to use the selected features to retrieve sentences from the corresponding item reviews [4].”
• NPA: Neural News Recommendation with Personalized Attention (2019, methodology): “• CNN [15], applying CNN to the word sequences in news titles and use max pooling technique to keep the most salient features, which is widely used in content-based recommendation [4, 40].”