Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Exploring the Individuality and Collectivity of Intents behind Interactions for Graph Collaborative Filtering

Yi Zhang, Lei Sang, Yiwen Zhang

TL;DR

The paper addresses the challenge of modeling fine-grained user-item intents in implicit-feedback recommender systems. It introduces BIGCF, a Bilateral Intent-guided Graph Collaborative Filtering framework that treats user and item motivations as collective and individualized intents and represents them with Gaussian latent embeddings in a graph-generation setting. Key contributions include a high-order graph structure encoder, bilateral intent-guided graph reconstruction, and augmentation-free graph contrastive regularization in both interaction and intent spaces, all trained end-to-end under an ELBO objective with reconstruction and regularization terms. Empirical results across Gowalla, Amazon-Book, and Tmall show BIGCF consistently outperforms baselines with robustness to data sparsity, supported by ablations and case studies that validate the importance of each component. The work advances practical recommender systems by combining causal intent modeling with scalable, self-supervised regularization to reveal users’ true preferences and improve recommendation quality.

Abstract

Intent modeling has attracted widespread attention in recommender systems. As the core motivation behind user selection of items, intent is crucial for elucidating recommendation results. The current mainstream modeling method is to abstract the intent into unknowable but learnable shared or non-shared parameters. Despite considerable progress, we argue that it still confronts the following challenges: firstly, these methods only capture the coarse-grained aspects of intent, ignoring the fact that user-item interactions will be affected by collective and individual factors (e.g., a user may choose a movie because of its high box office or because of his own unique preferences); secondly, modeling believable intent is severely hampered by implicit feedback, which is incredibly sparse and devoid of true semantics. To address these challenges, we propose a novel recommendation framework designated as Bilateral Intent-guided Graph Collaborative Filtering (BIGCF). Specifically, we take a closer look at user-item interactions from a causal perspective and put forth the concepts of individual intent-which signifies private preferences-and collective intent-which denotes overall awareness. To counter the sparsity of implicit feedback, the feature distributions of users and items are encoded via a Gaussian-based graph generation strategy, and we implement the recommendation process through bilateral intent-guided graph reconstruction re-sampling. Finally, we propose graph contrastive regularization for both interaction and intent spaces to uniformize users, items, intents, and interactions in a self-supervised and non-augmented paradigm. Experimental results on three real-world datasets demonstrate the effectiveness of BIGCF compared with existing solutions.

Exploring the Individuality and Collectivity of Intents behind Interactions for Graph Collaborative Filtering

TL;DR

The paper addresses the challenge of modeling fine-grained user-item intents in implicit-feedback recommender systems. It introduces BIGCF, a Bilateral Intent-guided Graph Collaborative Filtering framework that treats user and item motivations as collective and individualized intents and represents them with Gaussian latent embeddings in a graph-generation setting. Key contributions include a high-order graph structure encoder, bilateral intent-guided graph reconstruction, and augmentation-free graph contrastive regularization in both interaction and intent spaces, all trained end-to-end under an ELBO objective with reconstruction and regularization terms. Empirical results across Gowalla, Amazon-Book, and Tmall show BIGCF consistently outperforms baselines with robustness to data sparsity, supported by ablations and case studies that validate the importance of each component. The work advances practical recommender systems by combining causal intent modeling with scalable, self-supervised regularization to reveal users’ true preferences and improve recommendation quality.

Abstract

Intent modeling has attracted widespread attention in recommender systems. As the core motivation behind user selection of items, intent is crucial for elucidating recommendation results. The current mainstream modeling method is to abstract the intent into unknowable but learnable shared or non-shared parameters. Despite considerable progress, we argue that it still confronts the following challenges: firstly, these methods only capture the coarse-grained aspects of intent, ignoring the fact that user-item interactions will be affected by collective and individual factors (e.g., a user may choose a movie because of its high box office or because of his own unique preferences); secondly, modeling believable intent is severely hampered by implicit feedback, which is incredibly sparse and devoid of true semantics. To address these challenges, we propose a novel recommendation framework designated as Bilateral Intent-guided Graph Collaborative Filtering (BIGCF). Specifically, we take a closer look at user-item interactions from a causal perspective and put forth the concepts of individual intent-which signifies private preferences-and collective intent-which denotes overall awareness. To counter the sparsity of implicit feedback, the feature distributions of users and items are encoded via a Gaussian-based graph generation strategy, and we implement the recommendation process through bilateral intent-guided graph reconstruction re-sampling. Finally, we propose graph contrastive regularization for both interaction and intent spaces to uniformize users, items, intents, and interactions in a self-supervised and non-augmented paradigm. Experimental results on three real-world datasets demonstrate the effectiveness of BIGCF compared with existing solutions.
Paper Structure (25 sections, 20 equations, 7 figures, 5 tables)

This paper contains 25 sections, 20 equations, 7 figures, 5 tables.

Table of Contents

  1. Introduction
  2. METHODOLOGY
  3. Problem Formulation
  4. BIGCF
  5. High-order Graph Structure Encoding
  6. Bilateral Intent-guided Graph Reconstruction
  7. Graph Contrastive Regularization in Dual Spaces
  8. Multi-task Joint Training
  9. Model Analysis
  10. Time Complexity
  11. Theoretical Analysis
  12. Experiments
  13. Experimental Settings
  14. Datasets
  15. Baselines
  16. ...and 10 more sections

Figures (7)

  • Figure 1: (a) A simplified user-item interaction bipartite graph in movie recommendation scenario; (b) disentangled interaction graph incorporating user intents, user $u_1$'s choice of film $i_3$ is influenced by intents $c_1$ and $c_2$, indicating that he likes horror films with action elements; (c) user-item causal graph considering collective and individual attributes.
  • Figure 2: The complete framework of the proposed BIGCF. BIGCF consists of a high-order graph structure encoding, a bi-directional intent-directed graph reconstruction and a graph contrastive regularization process in dual space.
  • Figure 3: The computational process and information flow of graph contrastive regularization in interaction space and intent space. The solid line indicates that the input is directly involved in the computation, while the dashed line indicates that this input is indirectly involved in the operation.
  • Figure 4: Comparison of training time (seconds) per epoch for disentanglement-based methods on three datasets.
  • Figure 5: Sparsity tests with three interaction sparsity levels (sparse, normal, and popular) on (a) Gowalla, (b) Amazon-Book, and (c) Tmall datasets w.r.t. NDCG@20.
  • ...and 2 more figures