TEM: Tree-enhanced Embedding Model for Explainable Recommendation

By Xiang Wang, Xiangnan He, Fuli Feng, Liqiang Nie, Tat-seng Chua. 2018

259 citations recorded by the source corpus.

While collaborative filtering is the dominant technique in personalized recommendation, it models user-item interactions only and cannot provide concrete reasons for a recommendation. Meanwhile, the rich side information affiliated with user-item interactions (e.g., user demographics and item attributes), which provide valuable evidence that why a recommendation is suitable for a user, has not been fully explored in providing explanations. On the technical side, embedding-based methods, such as Wide&Deep and neural factorization machines, provide state-of-the-art recommendation performance. However, they work like a black-box, for which the reasons underlying a prediction cannot be explicitly presented. On the other hand, tree-based methods like decision trees predict by inferring decision rules from data. While being explainable, they cannot generalize to unseen feature interactions thus fail in collaborative filtering applications. In this work, we propose a novel solution named Tree-enhanced Embedding Method that combines the strengths of embedding-based and tree-based models. We first employ a tree-based model to learn explicit decision rules (aka. cross features) from the rich side information. We next design an embedding model that can incorporate explicit cross features and generalize to unseen cross features on user ID and item ID. At the core of our embedding method is an easy-to-interpret attention network, making the recommendation process fully transparent and explainable. We conduct experiments on two datasets of tourist attraction and restaurant recommendation, demonstrating the superior performance and explainability of our solution.

Subfields: Artificial Intelligence, Information Systems, Management Science and Operations Research

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. Tat-seng Chua
3. Xiang Wang
4. Liqiang Nie
5. Meng Wang
6. Yongfeng Zhang
7. Xu Chen
8. Fuli Feng

In Information Systems

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

In Management Science and Operations Research

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

Citations in the corpus

• Explainable Recommendation: A Survey and New Perspectives (2018, background): “Because user preferences on item features are dynamic and may change over time, Zhang et al. (2015b) extended the idea by modeling the user’s favorite features dynamically on daily resolution.”
• Unifying Knowledge Graph Learning and Recommendation: Towards a Better Understanding of User Preferences (2019, background): “This has been found to be important for the effectiveness, efficiency, persuasiveness, and user satisfaction of recommender systems [8, 39, 49].”
• EAPT: Efficient Attention Pyramid Transformer for Image Processing (2021, result): “include more feature information, and this has been observed in many previous works [23]–[25].”
• 3D-FUTURE: 3D Furniture Shape with TextURE (2020, methodology): “We also fine-tune the trained Pixel2Mesh here on the ShapeNetCore dataset.”
• A Reinforcement Learning Framework for Explainable Recommendation (2018, background): “Instead of refining existing recommendation models, researchers have also built inherently more explainable models by using intuitive data structures such as graphs [10], decision trees [31], and knowledge bases [32], [33].”
• Incorporating Interpretability into Latent Factor Models via Fast Influence Analysis (2018, unknown): “Wang et al. (2018) and Cheng et al. (2018b) empowered LFMs with explanations using both product features and review aspects.”
• Attentive Moment Retrieval in Videos (2018, background): “To tackle the aforementioned problem, we consider to explicitly capture the varying importance of each context moments by assigning an attentive weight to the embedding of each moment [37].”
• AutoInt: Automatic Feature Interaction Learning via Self-Attentive Neural Networks (2018, methodology): “Second, some tree-based methods [39, 42, 44] combined the power of embedding-based models and tree-based models but had to break training procedure into multiple stages.”
• A Survey on Accuracy-Oriented Neural Recommendation: From Collaborative Filtering to Information-Rich Recommendation (2021, methodology): “Furthermore, by combining embedding and tree based models seamlessly, TEM is able to unify their strengths — strong representation ability and explainability.”
• KGAT: Knowledge Graph Attention Network for Recommendation (2019, background): “data, much research effort has been devoted to collaborative filtering (CF) [12, 13, 32]. Despite its effectiveness and universality, CF methods suffer from the inability of modeling side information [30, 31], such as item attributes, user profiles, and contexts, thus perform poorly in sparse situat”
• Explainable Recommendation through Attentive Multi-View Learning (2019, methodology): “Explainability of the model, which is the ability to provide explanations for why an item is recommended, is considered equally important (Wang et al. 2018a).”
• LARA: Attribute-to-feature Adversarial Learning for New-item Recommendation (2020, methodology): “Recently, several deep learning models [16, 18, 36] have been applied to improve the recommendation performance via computing scores of the pairs of users and items.”
• Generate Natural Language Explanations for Recommendation (2021, background): “As a result, researchers have looked into explainable recommendation systems in the recent years [2, 3, 8, 16, 27, 38, 39, 44, 45], which can not only provide users with the recommendation lists, but also intutive explanations about why these items are recommended.”
• A historical perspective of explainable Artificial Intelligence (2020, methodology): “For each of the perspectives, the reader can find more comprehensive literature reviews in machine learning and Deep Learning (Arrieta et al., 2020; Fernandez, Herrera, Cordon, Jose del Jesus, & Marcelloni, 2019; Guidotti et al., 2018; Mueller, Hoffman, Clancey, Emrey, & Klein, 2019), recommender sy”
• Neural Compatibility Modeling with Attentive Knowledge Distillation (2018, methodology): “adopt the attention mechanism [1], which has proven to be effective in many machine learning tasks such as the recommendation [3, 6, 39] and representation learning [31].”
• Co-Attentive Multi-Task Learning for Explainable Recommendation (2019, methodology): “Since explainability is not included in the optimization goal, it is difficult to guarantee the quality of the explanations, e.g., consistency [Wang et al., 2018b].”