Hierarchical Cluster-based Open-World Graph Active Learning
sigir26-219 · Full research paper · Yayong Li, Zhengyi Du, Hong Zhang, Jonathan Wilton, Jinran Wu, Z L Liu, Nan Ye
Graph-based User and Item Representa · 100%
Abstract
Existing works in graph active learning (GAL) mostly assume a closed-world setting where all classes are known in advance, while in practice, we need to deal with the open-world setting where novel classes are encountered during learning. Apart from selecting informative nodes for refining the current classifier, the open-world GAL algorithms are also expected to discover novel classes. Motivated by the observation that identifying an informative region can be easier than finding the most informative example, we propose a novel hierarchical cluster-based algorithm for open-world GAL. Our algorithm performs clustering in the feature space to identify an informative region which either contains informative examples for known classes or novel classes, then performs a novel semi-supervised sub-clustering on the selected cluster to identify the most informative example. To facilitate the discovery of informative regions, we introduce a cluster-based self-distillation loss between ego and final embeddings to learn well-clustered node embeddings that are more likely to align with the classes. We also introduce a novel cluster informativeness score for the clusters, which measures not only how uncertain the cluster is (similar to standard active learning algorithms), but whether the cluster is likely to contain novel classes. Our semi-supervised sub-clustering algorithm partitions the selected cluster into subregions for known classes and an additional uncertain subregion, then we query the label for a representative node for the uncertain subregion. Extensive experiments on five benchmark datasets demonstrate the effectiveness of the proposed method with a more balanced improvement over novel classes. Ablation study shows that our cluster-based self-distillation loss, informativeness score, and the semi-supervised sub-clustering strategy are all beneficial. Empirical analysis also reveals how our informativeness score is effective for novel class discovery and identifying informative examples for refining the decision boundary.
Assigned reviewers
| # | Reviewer | Match | Why (closest evidence publications in the taxonomy) | |
|---|---|---|---|---|
| 1 | Xubin Ren | 47 | 100 How Expressive are Graph Neural Networks in Recommendation? (2023) Graph-based User and Item Representation Learning 77 Disentangled Contrastive Collaborative Filtering (2023) Graph Contrastive and Self-Supervised Learning 53 SSLRec: A Self-Supervised Learning Framework for Recommendation (2024) Graph Contrastive and Self-Supervised Learning |
|
| 2 | Yixin Liu 0001 | 45 | 100 Noise-Resilient Unsupervised Graph Representation Learning via Multi-Hop Feature Quality Estimation (2024) Graph-based User and Item Representation Learning 53 Beyond Smoothing: Unsupervised Graph Representation Learning with Edge Heterophily Discriminating (2023) Graph Contrastive and Self-Supervised Learning 53 Learning Strong Graph Neural Networks with Weak Information (2023) Graph Contrastive and Self-Supervised Learning |
Area chair
| Area chair | Match | |
|---|---|---|
| Qiaoyu Tan | 39 |
Area-chair candidates (4)
| Qiaoyu Tan assigned | 39 | 7 papers | |
| Weiming Liu 0005 | 37 | 8 papers | |
| Leyu Lin | 35 | 15 papers | |
| Yanan Niu | 34 | 15 papers |
Candidates from the panel ranked by match score
| # | Reviewer | Match | Load | Why | |
|---|---|---|---|---|---|
| 1 | Jing Zhu 0005 | 48 | 3 | 100 Node Proximity Is All You Need: Unified Structural and Positional Node and Graph Embedding (2021) Graph-based User and Item Representation Learning 100 CAPER: Coarsen, Align, Project, Refine - A General Multilevel Framework for Network Alignment (2022) Graph-based User and Item Representation Learning |
|
| 2 | Xubin Ren assigned | 47 | 3 | 100 How Expressive are Graph Neural Networks in Recommendation? (2023) Graph-based User and Item Representation Learning 77 Disentangled Contrastive Collaborative Filtering (2023) Graph Contrastive and Self-Supervised Learning |
|
| 3 | Mengyuan Jing | 47 | 3 | 100 Learning Shared Representations for Recommendation with Dynamic Heterogeneous Graph Convolutional Networks (2023) Graph-based User and Item Representation Learning 53 Graph Contrastive Learning with Adaptive Augmentation for Recommendation (2022) Graph Contrastive and Self-Supervised Learning |
|
| 4 | Yixin Liu 0001 assigned | 45 | 3 | 100 Noise-Resilient Unsupervised Graph Representation Learning via Multi-Hop Feature Quality Estimation (2024) Graph-based User and Item Representation Learning 53 Beyond Smoothing: Unsupervised Graph Representation Learning with Edge Heterophily Discriminating (2023) Graph Contrastive and Self-Supervised Learning |
|
| 5 | Lianghao Xia | 45 | 3 | 53 Graph Augmentation for Recommendation (2024) Graph Contrastive and Self-Supervised Learning 53 Graph Transformer for Recommendation (2023) Graph Contrastive and Self-Supervised Learning |
|
| 6 | Xurong Liang | 44 | 3 | 63 Lightweight Embeddings for Graph Collaborative Filtering (2024) Graph-based User and Item Representation Learning 63 Lightweight Embeddings with Graph Rewiring for Collaborative Filtering (2025) Graph-based User and Item Representation Learning |
|
| 7 | Yi Zhang 0103 | 44 | 3 | 66 Revisiting Graph-based Recommender Systems from the Perspective of Variational Auto-Encoder (2023) Graph-based User and Item Representation Learning 65 AdaGIN: Adaptive Graph Interaction Network for Click-Through Rate Prediction (2025) CTR and Conversion Rate Prediction |
|
| 8 | Dongxiao He | 44 | 3 | 53 Contrastive Learning Meets Homophily: Two Birds with One Stone (2023) Graph Contrastive and Self-Supervised Learning 53 Self-Supervised Graph Neural Networks via Diverse and Interactive Message Passing (2022) Graph Contrastive and Self-Supervised Learning |
|
| 9 | Xiaolong Liu 0012 | 43 | 3 | 76 Graph-based Alignment and Uniformity for Recommendation (2023) Graph-based User and Item Representation Learning 67 DGRec: Graph Neural Network for Recommendation with Diversified Embedding Generation (2023) Graph-based User and Item Representation Learning |
|
| 10 | Chen Wang 0052 | 43 | 3 | 76 Graph-based Alignment and Uniformity for Recommendation (2023) Graph-based User and Item Representation Learning 52 Instruction-based Hypergraph Pretraining (2024) Graph Pretraining and Foundation Models |
|
| 11 | Ke Wang 0038 | 42 | 3 | 52 Contrastive Self-supervised Learning in Recommender Systems: A Survey (2024) Graph Contrastive and Self-Supervised Learning 48 Disentangled Contrastive Learning for Cross-Domain Recommendation (2023) Graph Contrastive and Self-Supervised Learning |
|
| 12 | Hyunsik Yoo | 41 | 3 | 100 Directed Network Embedding with Virtual Negative Edges (2022) Graph-based User and Item Representation Learning 70 Disentangling Degree-related Biases and Interest for Out-of-Distribution Generalized Directed Network Embedding (2023) Graph Anomaly and Out-of-Distribution Detection |
|
| 13 | Jurek Leonhardt | 41 | 3 | 100 Node Representation Learning for Directed Graphs (2019) Graph-based User and Item Representation Learning 38 Data Augmentation for Sample Efficient and Robust Document Ranking (2024) Ranking Distillation and Adaptation |
|
| 14 | Wei Wei 0027 | 40 | 3 | 56 LLMRec: Large Language Models with Graph Augmentation for Recommendation (2024) Graph Pretraining and Foundation Models 51 Heterogeneous Graph Contrastive Learning for Recommendation (2023) Graph Contrastive and Self-Supervised Learning |
|
| 15 | Chunyang Wang 0001 | 40 | 3 | 56 Learning Graph-based Disentangled Representations for Next POI Recommendation (2022) Sequential and Session-Based Graph Recommendation 47 Review-Enhanced Hierarchical Contrastive Learning for Recommendation (2024) Graph Contrastive and Self-Supervised Learning |
|
| 17 | Yongjing Hao | 39 | 3 | 67 How Do Graph Signals Affect Recommendation: Unveiling the Mystery of Low and High-Frequency Graph Signals (2026) Graph-based User and Item Representation Learning 48 Meta-optimized Structural and Semantic Contrastive Learning for Graph Collaborative Filtering (2024) Graph Contrastive and Self-Supervised Learning |
|
| 18 | Mingdai Yang | 39 | 3 | 76 Graph-based Alignment and Uniformity for Recommendation (2023) Graph-based User and Item Representation Learning 52 Instruction-based Hypergraph Pretraining (2024) Graph Pretraining and Foundation Models |
|
| 19 | Yinwei Wei | 39 | 3 | 51 MultiCBR: Multi-view Contrastive Learning for Bundle Recommendation (2024) Graph Contrastive and Self-Supervised Learning 46 MMGCN: Multi-modal Graph Convolution Network for Personalized Recommendation of Micro-video (2019) Graph Contrastive and Self-Supervised Learning |
|
| 20 | Jiuqiang Li 0002 | 38 | 3 | 44 Graph Diffusive Self-Supervised Learning for Social Recommendation (2024) Social and Group Recommendation 37 Disentangled Contrastive Bundle Recommendation with Conditional Diffusion (2025) Diffusion and Flow Recommenders |
|
| 21 | Jiancan Wu | 37 | 3 | 67 Graph convolution machine for context-aware recommender system (2022) Graph-based User and Item Representation Learning 53 Self-supervised Graph Learning for Recommendation (2021) Graph Contrastive and Self-Supervised Learning |
|
| 23 | Bowen Zheng 0005 | 37 | 3 | 53 Enhancing Graph Contrastive Learning with Reliable and Informative Augmentation for Recommendation (2025) Graph Contrastive and Self-Supervised Learning 40 Adapting Large Language Models by Integrating Collaborative Semantics for Recommendation (2024) Generative Recommendation and Item Tokenization |
|
| 24 | Jinheon Baek | 37 | 3 | 100 Edge Representation Learning with Hypergraphs (2021) Graph-based User and Item Representation Learning 100 Accurate Learning of Graph Representations with Graph Multiset Pooling (2021) Graph-based User and Item Representation Learning |
|
| 25 | Lingling Yi | 36 | 3 | 63 PlatoGL: Effective and Scalable Deep Graph Learning System for Graph-enhanced Real-Time Recommendation (2022) Graph-based User and Item Representation Learning 47 Dual Interests-Aligned Graph Auto-Encoders for Cross-domain Recommendation in WeChat (2023) Graph Contrastive and Self-Supervised Learning |
|
| 26 | Jianghong Ma | 36 | 3 | 64 ExpGCN: Review-aware Graph Convolution Network for explainable recommendation (2023) Graph-based User and Item Representation Learning 63 Diversity Recommendation via Causal Deconfounding of Co-purchase Relations and Counterfactual Exposure (2026) Graph-based User and Item Representation Learning |
|
| 27 | Fei Sun 0001 | 36 | 3 | 64 Exact-K Recommendation via Maximal Clique Optimization (2019) Graph-based User and Item Representation Learning 44 BERT4Rec: Sequential Recommendation with Bidirectional Encoder Representations from Transformer (2019) Graph-Based Sequential and Session Recommendation |
|
| 28 | Lantao Hu | 35 | 3 | 53 Towards Robust Recommendation via Decision Boundary-aware Graph Contrastive Learning (2024) Graph Contrastive and Self-Supervised Learning 45 Soft Contrastive Sequential Recommendation (2024) Contrastive and Self-Supervised Sequential Recommendation |
|
| 29 | Noveen Sachdeva | 35 | 3 | 100 GC4NC: A Benchmark Framework for Graph Condensation on Node Classification with New Insights (2025) Graph-based User and Item Representation Learning 38 ActionPiece: Contextually Tokenizing Action Sequences for Generative Recommendation (2025) Item Tokenization and Semantic IDs |
|
| 31 | Jianling Wang | 35 | 3 | 45 Session-based Recommendation with Hypergraph Attention Networks (2021) Sequential and Session-Based Graph Recommendation 45 Next-item Recommendation with Sequential Hypergraphs (2020) Sequential and Session-Based Graph Recommendation |
|
| 32 | Bo Peng 0009 | 35 | 3 | 44 Modeling Sequences as Star Graphs to Address Over-Smoothing in Self-Attentive Sequential Recommendation (2024) Graph-Based Sequential and Session Recommendation 44 Intention enhanced mixed attentive model for session-based recommendation (2024) Graph-Based Sequential and Session Recommendation |
|
| 33 | Zhengyi Yang 0007 | 35 | 3 | 40 Large Language Model Can Interpret Latent Space of Sequential Recommender (2026) Generative Recommendation and Item Tokenization 39 A Bi-Step Grounding Paradigm for Large Language Models in Recommendation Systems (2025) Generative Recommendation and Item Tokenization |