Learning Disentangled Representations for Recommendation

By Jianxin Ma, Chang Zhou, Peng Cui, Hongxia Yang, Wenwu Zhu. 2019

399 citations recorded by the source corpus.

User behavior data in recommender systems are driven by the complex interactions of many latent factors behind the users’ decision making processes. The factors are highly entangled, and may range from high-level ones that govern user intentions, to low-level ones that characterize a user’s preference when executing an intention. Learning representations that uncover and disentangle these latent factors can bring enhanced robustness, interpretability, and controllability. However, learning such disentangled representations from user behavior is challenging, and remains largely neglected by the existing literature. In this paper, we present the MACRo-mIcro Disentangled Variational Auto-Encoder (MacridVAE) for learning disentangled representations from user behavior. Our approach achieves macro disentanglement by inferring the high-level concepts associated with user intentions (e.g., to buy a shirt or a cellphone), while capturing the preference of a user regarding the different concepts separately. A micro-disentanglement regularizer, stemming from an information-theoretic interpretation of VAEs, then forces each dimension of the representations to independently reflect an isolated low-level factor (e.g., the size or the color of a shirt). Empirical results show that our approach can achieve substantial improvement over the state-of-the-art baselines. We further demonstrate that the learned representations are interpretable and controllable, which can potentially lead to a new paradigm for recommendation where users are given fine-grained control over targeted aspects of the recommendation lists.

Subfields: Computer Vision and Pattern Recognition, Experimental and Cognitive Psychology, 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 Computer Vision and Pattern Recognition

1. Xiangnan He
2. Xiang Wang
3. Tat-seng Chua
4. Fuli Feng
5. Xinchao Wang
6. Yong Li
7. Chen Gao
8. Chao Huang

In Experimental and Cognitive Psychology

1. Xiangnan He

In Information Systems

1. Xiangnan He
2. Xiang Wang
3. Tat-seng Chua
4. Hongzhi Yin
5. Fuli Feng
6. Chao Huang
7. Yong Li
8. Chen Gao

Citations in the corpus

• Learning Intents behind Interactions with Knowledge Graph for Recommendation (2021, background): “Different intents should contain different information about user preference [23, 24].”
• Robust Federated Recommendation System (2020, background): “The evaluation metric are Precision@ K [38] and Recall@ K [22] with the ranking position K ranges from 1 to 5 .”
• Factorizable Graph Convolutional Networks (2020, methodology): “Besides designing GCN to better extract information from non-grid data, there are also a couple of works that explore the disentangled GCNs [Ma et al., 2019b, Liu et al., 2019].”
• CausalVAE: Disentangled Representation Learning via Neural Structural Causal Models (2020, background): “Disentangled representation learning is of great importance in various applications such as computer vision, speech and natural language processing, and recommender systems [9, 20, 8].”
• Controllable Multi-Interest Framework for Recommendation (2020, methodology): “We evaluate the performance of all methods under strong generalization [34, 37, 38]: We split all users into training/validation/test sets by the proportion of 8:1:1.”
• Disentangled Graph Collaborative Filtering (2020, methodology): “We conduct extensive experiments on three benchmark datasets, and DGCF achieves significant improvements over several state-of-the-art models like NGCF [40], DisenGCN [25], and MacridVAE [26].”
• Disentangling User Interest and Conformity for Recommendation with Causal Embedding (2021, methodology): “Disentangled representation learning in recommender systems was not explored until recently [14, 33, 47].”
• Deep Learning on Graphs: A Survey (2018, neutral): “h as node or graph classification10, graph-based deep learning methods have also been applied to a wide range of disciplines, including modeling social influence [133], recommendation [28], [66], [99], [134], chemistry and biology [46], [52], [55], [116], [117], physics [135], [136], disease and drug ”
• Self-Supervised Learning for Recommender Systems: A Survey (2022, methodology): “Similar tasks are also used for user preference disentanglement [130].”
• Diffusion Recommender Model (2023, methodology): “Existing work on generative recommender models can be roughly divided into two groups: GAN-based methods [7, 52] and VAE-based methods [23, 26].”
• Adversarial and Contrastive Variational Autoencoder for Sequential Recommendation (2021, positive): “MacridVAE [15] employs VAE to learn disentangled representations that can enhance robustness.”
• Graph Neural Networks in Recommender Systems: A Survey (2020, background): “, disentangled representation learning [105, 106] or capsule networks [82].”
• Hypergraph Contrastive Collaborative Filtering (2022, background): “, DGCF [36], MacridVAE [24] and MBGMN [41].”
• RecBole: Towards a Unified, Comprehensive and Efficient Framework for Recommendation Algorithms (2020, neutral): “MacridVAE [41] NeurIPS 2019 AliShop-7C, MovieLens, Netflix”
• A Survey on Heterogeneous Graph Embedding: Methods, Techniques, Applications and Sources (2020, neutral): “tangled learning [155], [156], which divides the embedding into different latent spaces to improve the interpretability, can also be considered.”