Learning Noise-Resilient and Transferable Graph-Text Alignment via Dynamic Quality Assessment
sigir26-055 · Full research paper · Yuhang Liu, Minglai Shao, Zengyi Wo, Yunlong Chu, Bing Hao, Shengzhong Liu, Ruijie Wang, Jianxin Li
Entity Linking and Alignment · 50% Graph Pretraining and Foundation Mod · 25% Graph Learning for Text-Attributed a · 25%
Abstract
Pre-training graph foundation models (GFMs) on text-attributed graphs (TAGs) is important for web-scale retrieval and recommendation, where graph entities are matched with textual descriptions. Existing CLIP-style graph-text aligners typically assume one-to-one correspondence: each node is pulled close only to its paired text, and all other pairs are treated as negatives. This overlooks the many-to-many relations common in real TAGs, where a node and its local neighborhood can be semantically related to multiple texts, and vice versa. Meanwhile, TAG supervision is often imperfect: noisy or weak node-text links introduce false-positive pairs, causing contrastive learning to align mismatched semantics. These limitations reveal a fundamental trade-off: leveraging expressive many-to-many signals increases semantic coverage but may propagate errors under noise, whereas strict one-to-one training is more conservative yet still suffers when mismatched pairs remain in the training set. Therefore, we propose ADAligner, a quality-aware graph–text alignment framework that adapts between expressive many-to-many and conservative one-to-one objectives based on estimated alignment reliability. ADAligner tracks batch-level reliability online and adjusts optimization accordingly—promoting soft, subgraph-level alignment when supervision is clean while emphasizing reliable one-to-one alignment by filtering low-confidence pairs under noise. We provide theoretical analysis showing that this closed-loop adaptation is stable and convergent. Experiments on nine TAG benchmarks show that, under 30% mismatched node-text supervision, ADAligner consistently improves cross-modal retrieval by 144.70% on average, zero-/few-shot node classification by 26.13%, and link prediction by 4.70% over the strongest multimodal baseline, demonstrating strong robustness to alignment noise across both unsupervised and transfer settings. Our code is available at https://github.com/karmaisacat-13/ADAligner.
Assigned reviewers
| # | Reviewer | Match | Why (closest evidence publications in the taxonomy) | |
|---|---|---|---|---|
| 1 | Yuzhong Qu | 62 | 67 Revisiting Embedding-Based Entity Alignment: A Robust and Adaptive Method (2023) Entity Linking and Alignment 67 Facilitating Entity Navigation Through Top-K Link Patterns (2015) Entity Linking and Alignment 67 Summarizing Entity Descriptions for Effective and Efficient Human-centered Entity Linking (2015) Entity Linking and Alignment |
|
| 2 | Jing Zhu 0005 | 55 | 70 LinkGPT: Leveraging Large Language Models for Enhanced Link Prediction in Text-Attributed Graphs (2025) Knowledge Graph Embedding and Completion 67 TouchUp-G: Improving Feature Representation through Graph-Centric Finetuning (2024) Graph Pretraining and Foundation Models 56 Pitfalls in Link Prediction with Graph Neural Networks: Understanding the Impact of Target-link Inclusion & Better Practices (2024) Graph Learning for Text-Attributed and Heterogeneous Networks |
|
| 3 | Jingwei Cheng | 47 | 67 Probing Relative Interaction and Dynamic Calibration in Multi-modal Entity Alignment (2025) Entity Linking and Alignment 67 SGMEA: Structure-Guided Multimodal Entity Alignment (2025) Entity Linking and Alignment 67 Breaking the Noise Barrier: LLM-Guided Semantic Filtering and Enhancement for Multi-Modal Entity Alignment (2025) Entity Linking and Alignment |
Area chair
| Area chair | Match | |
|---|---|---|
| Qiaoyu Tan | 37 |
Area-chair candidates (4)
| Qiaoyu Tan assigned | 37 | 7 papers | |
| Chenyan Xiong | 35 | 15 papers | |
| Xiaoxue Zang | 32 | 4 papers | |
| Kam-Fai Wong COI: co-author | 32 | 6 papers |
Candidates from the panel ranked by match score
| # | Reviewer | Match | Load | Why | |
|---|---|---|---|---|---|
| 1 | Yuzhong Qu assigned | 62 | 3 | 67 Revisiting Embedding-Based Entity Alignment: A Robust and Adaptive Method (2023) Entity Linking and Alignment 67 Facilitating Entity Navigation Through Top-K Link Patterns (2015) Entity Linking and Alignment |
|
| 2 | Jing Zhu 0005 assigned | 55 | 3 | 70 LinkGPT: Leveraging Large Language Models for Enhanced Link Prediction in Text-Attributed Graphs (2025) Knowledge Graph Embedding and Completion 67 TouchUp-G: Improving Feature Representation through Graph-Centric Finetuning (2024) Graph Pretraining and Foundation Models |
|
| 3 | Dongxiao He COI: co-author | 54 | 3 | 67 DuoKD: Dual Knowledge Distillation from Large Language Models for Robust Graph Neural Networks (2026) Graph Pretraining and Foundation Models 62 LLM-FSGNN: LLM-Guided Feature-Structure Augmentation Graph Neural Network for Cold-Start Recommendation (2026) LLM-Derived Knowledge Infusion |
|
| 4 | Fu Zhang 0001 | 50 | 3 | 67 Probing Relative Interaction and Dynamic Calibration in Multi-modal Entity Alignment (2025) Entity Linking and Alignment 67 SGMEA: Structure-Guided Multimodal Entity Alignment (2025) Entity Linking and Alignment |
|
| 5 | Jingwei Cheng assigned | 47 | 3 | 67 Probing Relative Interaction and Dynamic Calibration in Multi-modal Entity Alignment (2025) Entity Linking and Alignment 67 SGMEA: Structure-Guided Multimodal Entity Alignment (2025) Entity Linking and Alignment |
|
| 6 | Ben Liu 0002 | 46 | 3 | 52 Enhanced Expert Merging for Mixture-of-Experts in Graph Foundation Models (2025) Graph Pretraining and Foundation Models 42 Clarified Aggregation and Predictive Modeling (CAPM): High-Interpretability Framework for Inductive Link Prediction (2024) Knowledge Graph Embedding and Completion |
|
| 7 | Wonduk Seo | 46 | 3 | 67 Question-to-Knowledge (Q2K): Multi-Agent Generation of Inspectable Facts for Product Mapping (2025) Entity Linking and Alignment 30 A New Query Expansion Approach for Enhancing Information Retrieval via Agent-Mediated Dialogic Inquiry (2026) Query Reformulation and Expansion |
|
| 8 | Xuhui Sui | 45 | 3 | 67 Introducing high correlation and high quality instances for few-shot entity linking (2025) Entity Linking and Alignment 67 Improving Zero-Shot Entity Linking Candidate Generation with Ultra-Fine Entity Type Information (2022) Entity Linking and Alignment |
|
| 9 | Qinggang Zhang | 44 | 3 | 67 Entity Alignment with Noisy Annotations from Large Language Models (2024) Entity Linking and Alignment 62 LoSemB: Logic-Guided Semantic Bridging for Inductive Tool Retrieval (2026) Reasoning Evaluation and Benchmarks |
|
| 10 | Tianzhe Zhao | 43 | 3 | 58 Incorporating Context Graph with Logical Reasoning for Inductive Relation Prediction (2022) Knowledge Graph Embedding and Completion 58 Inductive Relation Prediction with Logical Reasoning Using Contrastive Representations (2022) Knowledge Graph Embedding and Completion |
|
| 11 | Xubin Ren | 43 | 3 | 52 A Survey of Large Language Models for Graphs (2024) Graph Pretraining and Foundation Models 47 Graph Transformer for Recommendation (2023) Graph Contrastive and Self-Supervised Learning |
|
| 12 | Jiapu Wang | 41 | 3 | 42 TDN: Triplet Distributor Network for Knowledge Graph Completion (2023) Knowledge Graph Embedding and Completion 42 MADE: Multicurvature Adaptive Embedding for Temporal Knowledge Graph Completion (2024) Knowledge Graph Embedding and Completion |
|
| 13 | Xiaolong Jin 0001 | 41 | 3 | 42 Knowledge Graph Embedding: A Locally and Temporally Adaptive Translation-Based Approach (2018) Knowledge Graph Embedding and Completion 42 Locally Adaptive Translation for Knowledge Graph Embedding (2016) Knowledge Graph Embedding and Completion |
|
| 14 | Xiangyu Ke | 40 | 3 | 67 Select Your Questions Wisely: For Entity Resolution With Crowd Errors (2017) Entity Linking and Alignment 67 In-context Clustering-based Entity Resolution with Large Language Models: A Design Space Exploration (2025) Entity Linking and Alignment |
|
| 15 | Chen Wang 0052 | 40 | 3 | 62 Multi-task Item-attribute Graph Pre-training for Strict Cold-start Item Recommendation (2023) Graph Pretraining and Foundation Models 52 Instruction-based Hypergraph Pretraining (2024) Graph Pretraining and Foundation Models |
|
| 16 | Mingdai Yang | 40 | 3 | 52 Instruction-based Hypergraph Pretraining (2024) Graph Pretraining and Foundation Models 48 Enhanced Pre-training for Recommendation via Hypergraph Structural Entropy (2026) Graph Pretraining and Foundation Models |
|
| 17 | Lianghao Xia | 39 | 3 | 59 Knowledge Graph Self-Supervised Rationalization for Recommendation (2023) Knowledge Graph Embedding and Completion 52 GraphPro: Graph Pre-training and Prompt Learning for Recommendation (2024) Graph Pretraining and Foundation Models |
|
| 18 | Yixin Liu 0001 | 39 | 3 | 52 SpecG: A Spectral-Based Framework for Effective Graph Pretraining and Knowledge Transfer (2025) Graph Pretraining and Foundation Models 40 Noise-Resilient Unsupervised Graph Representation Learning via Multi-Hop Feature Quality Estimation (2024) Graph-based User and Item Representation Learning |
|
| 19 | Wenyu Huang | 38 | 3 | 59 A Large-scale Offer Alignment Model for Partitioning Filtering and Matching Product Offers (2024) Entity Linking and Alignment 35 Prompting large language models with knowledge graphs for question answering involving long-tail facts (2025) KG-Augmented LLM Reasoning |
|
| 20 | Armin Toroghi | 38 | 3 | 41 Bayesian Inference with Complex Knowledge Graph Evidence (2024) Knowledge Graph Question Answering 38 CommonWhy: A Dataset for Evaluating Entity-Based Causal Commonsense Reasoning in Large Language Models (2026) Knowledge Graph Question Answering |
|
| 21 | Yu Gu 0016 | 38 | 3 | 59 Reviving the Context: Camera Trap Species Classification as Link Prediction on Multimodal Knowledge Graphs (2024) Multimodal and Cross-Modal Retrieval › Others 42 Relevance Search over Schema-Rich Knowledge Graphs (2019) Knowledge Graph Embedding and Completion |
|
| 22 | Ke Wang 0038 | 37 | 3 | 51 Inter- and Intra-Domain Relation-Aware Heterogeneous Graph Convolutional Networks for Cross-Domain Recommendation (2022) Graph Contrastive and Self-Supervised Learning 50 Learning Aspect-Aware High-Order Representations from Ratings and Reviews for Recommendation (2023) Graph Learning for Text-Attributed and Heterogeneous Networks |
|
| 23 | Wei Wei 0027 | 37 | 3 | 54 Heterogeneous Graph Contrastive Learning for Recommendation (2023) Graph Contrastive and Self-Supervised Learning 49 LLMRec: Large Language Models with Graph Augmentation for Recommendation (2024) Graph Pretraining and Foundation Models |
|
| 25 | Xiaolong Liu 0012 | 37 | 3 | 52 Instruction-based Hypergraph Pretraining (2024) Graph Pretraining and Foundation Models 39 Graph-based Alignment and Uniformity for Recommendation (2023) Graph-based User and Item Representation Learning |
|
| 26 | Jinheon Baek | 36 | 3 | 58 Learning to Extrapolate Knowledge: Transductive Few-shot Out-of-Graph Link Prediction (2020) Knowledge Graph Embedding and Completion 40 Accurate Learning of Graph Representations with Graph Multiset Pooling (2021) Graph-based User and Item Representation Learning |
|
| 27 | Xinying Qian | 35 | 3 | 42 Contrast then Memorize: Semantic Neighbor Retrieval-Enhanced Inductive Multimodal Knowledge Graph Completion (2024) Knowledge Graph Embedding and Completion 42 Bring Invariant to Variant: A Contrastive Prompt-based Framework for Temporal Knowledge Graph Forecasting (2024) Knowledge Graph Embedding and Completion |
|
| 29 | Yi Zhang 0103 | 34 | 3 | 54 Intent-Guided Heterogeneous Graph Contrastive Learning for Recommendation (2025) Graph Contrastive and Self-Supervised Learning 53 Heterogeneous Adaptive Preference Learning for Recommendation (2026) Graph Contrastive and Self-Supervised Learning |
|
| 30 | Yinwei Wei | 34 | 3 | 37 MultiCBR: Multi-view Contrastive Learning for Bundle Recommendation (2024) Graph Contrastive and Self-Supervised Learning 37 MMGCN: Multi-modal Graph Convolution Network for Personalized Recommendation of Micro-video (2019) Graph Contrastive and Self-Supervised Learning |
|
| 31 | Jurek Leonhardt | 33 | 3 | 40 Node Representation Learning for Directed Graphs (2019) Graph-based User and Item Representation Learning 34 Data Augmentation for Sample Efficient and Robust Document Ranking (2024) Ranking Distillation and Adaptation |
|
| 32 | Xingyu Tan 0001 | 33 | 3 | 38 PRoH: Dynamic Planning and Reasoning over Knowledge Hypergraphs for Retrieval-Augmented Generation (2026) Multimodal and Graph RAG 36 HydraRAG: Structured Cross-Source Enhanced Large Language Model Reasoning (2025) KG-Augmented LLM Reasoning |