TriRank: Review-aware Explainable Recommendation by Modeling Aspects

By Xiangnan He, Tao Chen, Min-Yen Kan, Xiao Chen. 2015

551 citations recorded by the source corpus.

Most existing collaborative filtering techniques have focused on modeling the binary relation of users to items by extracting from user ratings. Aside from users' ratings, their affiliated reviews often provide the rationale for their ratings and identify what aspects of the item they cared most about. We explore the rich evidence source of aspects in user reviews to improve top-N recommendation. By extracting aspects (i.e., the specific properties of items) from textual reviews, we enrich the user--item binary relation to a user--item--aspect ternary relation. We model the ternary relation as a heterogeneous tripartite graph, casting the recommendation task as one of vertex ranking. We devise a generic algorithm for ranking on tripartite graphs -- TriRank -- and specialize it for personalized recommendation. Experiments on two public review datasets show that it consistently outperforms state-of-the-art methods. Most importantly, TriRank endows the recommender system with a higher degree of explainability and transparency by modeling aspects in reviews. It allows users to interact with the system through their aspect preferences, assisting users in making informed decisions.

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. Yongfeng Zhang
3. Xu Chen
4. Tat-seng Chua
5. Xiang Wang
6. Liqiang Nie
7. Yi Tay
8. Xing Xie

In Information Systems

1. Xiangnan He
2. Yongfeng Zhang
3. Xu Chen
4. Tat-seng Chua
5. Xiang Wang
6. Liqiang Nie
7. Xing Xie
8. Fuli Feng

Citations in the corpus

• Explainable Recommendation: A Survey and New Perspectives (2018, methodology): “In early research stages, Herlocker et al. (2000), Bilgic and Mooney (2005), and Tintarev and Masthoff (2007b) and McSherry (2005) adopted statistical histograms or pie charts as explanations to help users understand the rating distribution and the pros/cons of a recommendation.”
• Neural Rating Regression with Abstractive Tips Generation for Recommendation (2017, methodology): “For user review texts, some research works, such as HFT [23], RMR [21], TriRank [10], and sCVR [26], integrate topic models in their frameworks to generate the latent factors for users and items incorporating review texts.”
• Visually Explainable Recommendation (2018, methodology): “By modeling aspects in user reviews, He et al [9] devised a graph algo-rithm called TriRank for providing recommendations with better explainability and transparency, and to leverage user opinions as well as social information, Ren et al [23] introduced the concept of viewpoints (a tuple of…”
• ReEL: Review Aware Explanation of Location Recommendation (2018, methodology): “[21] exploited tri-partite modeling of user-item-aspect tuples and used graph-based ranking to find the most relevant aspects of a user that match with relevant aspects of places.”
• Aspect Based Recommendations: Recommending Items with the Most Valuable Aspects Based on User Reviews (2017, methodology): “Further, [12] applies a vertex ranking approach to the tri-partite graph of users-items-aspects to provide beŠer recommendations of items using reviews.”
• Transparent, Scrutable and Explainable User Models for Personalized Recommendation (2019, background): “Various forms of explanations have been explored in prior work, including sentences [21, 55], tag/keyword clouds [17, 30, 53], as well as different kinds of visualizations [10, 22, 31, 47].”
• Distilling Structured Knowledge into Embeddings for Explainable and Accurate Recommendation (2019, methodology): “TriRank [8] and SULM [1] utilize sentiment analysis techniques to extract aspects and user opinions to produce recommendation explanations.”
• TEM: Tree-enhanced Embedding Model for Explainable Recommendation (2018, methodology): “[19] extracted item aspects from user reviews and modeled the user-item-aspect relations in a hybrid collaborative filtering model.”
• Interest-aware Message-Passing GCN for Recommendation (2021, methodology): “Two widely used evaluation metrics for top-𝑛 recommendation are adopted in our evaluation: Recall and Normalized Discounted Cumulative Gain [13].”
• Sequential Recommendation with User Memory Networks (2018, methodology): “Based on these discovered patterns, the item-level RUM can interpret recommender system from the sequential behavior perspective, which is different from previous methods that usually leverage side information (e.g. user textual reviews [2, 12, 33, 34]) for explanations.”
• Explainable Movie Recommendation Systems by using Story-based Similarity (2018, background): “There have been various studies [7, 2] on building ‘explainable’ recommender systems.”
• Co-Attentive Multi-Task Learning for Explainable Recommendation (2019, methodology): “Embedded methods integrate the explanation process into the construction of the recommendation model [McAuley and Leskovec, 2013; Zhang et al., 2014; He et al., 2015; Chen et al., 2018; Wang et al., 2018a].”
• Neural Attentional Rating Regression with Review-level Explanations (2018, background): “models in [11, 32, 44], as the pre-selection of reviews, which will be studied in the future.”
• Automatic Generation of Natural Language Explanations (2017, background): “On the other hand, in [8] they propose a tripartite graph to enrich the user-item binary relation to a user-item-aspect ternary relation.”
• DELF: A Dual-Embedding based Deep Latent Factor Model for Recommendation (2018, methodology): “We used Hit Ratio (HR) and Normalized Discounted Cumulative Gain (NDCG) [He et al. , 2015] as metrics.”