Aug 2026· Proceedings of the 32nd ACM SIGKDD Conference on Knowledge Discovery and Data Mining V.2· pp. 4754-4765· 0 citations· 25 references
Abstract
Dynamic text-attributed graphs (DyTAGs) exhibit coupled textual and structural dynamics, and existing mainstream approaches for DyTAGs extend conventional large language models (LLMs) to capture both dynamics, thereby giving rise to dynamic graph LLMs. However, in DyTAGs, the continuous emergence of new nodes and edges with incoming textual content and interactions drives the joint evolution of graph structural-textual patterns, causing existing methods to struggle with evolving patterns. This motivates a largely unexplored problem of continual learning on DyTAGs, which aims to adapt to constantly evolving graph structural-textual patterns while retaining past knowledge, which imposes two challenges: 1) unlike common graphs, graph structure and textual semantics in emerging DyTAG patterns jointly evolve, requiring dynamic graph LLMs to adapt structure, text, and graph-text fusion simultaneously; and 2) updating dynamic graph LLMs to fit a new pattern may destroy the global graph-text fusion capabilities and bias the model towards recent local dynamics. To address these challenges, we propose a novel Continual Learning Dynamic Graph LLM framework (Continual-GraphLLM) to continually adapt to incoming patterns by routing them to experts specialized in similar past patterns, while mitigating the overwriting of previously learned patterns by assigning new experts to unseen patterns. Specifically, we propose a graph-text factor-based router to adapt to incoming structural-textual joint patterns by utilizing latent factors to adaptively activate suitable experts. Furthermore, we design invariance regularized multi-scale experts that mitigate forgetting by capturing the invariances among learned patterns assigned to the same expert, where each expert progressively integrates structural and textual information from local scale to global scale. Extensive experiments on real-world DyTAGs demonstrate the superiority of our method over competitive baselines, highlighting its effectiveness in adapting to emerging DyTAG patterns.
Temporal graph learning models the evolution of dynamic systems, where both structural interactions and semantic states change over time. However, existing benchmarks primarily emphasize structural evolution via temporal link prediction (TLP), while support for semantic evolution remains limited. Although temporal node classification (TNC) is sometimes included, it is typically restricted to simplistic binary settings that fail to capture realistic semantic drift. Moreover, commonly used datasets exhibit high link repetition, leading to inflated performance estimates and obscuring true model capability. To address these limitations, we introduce \textbf{TTGBench}, a new benchmark that jointly evaluates structural and semantic evolution. TTGBench comprises six real-world, text-rich datasets characterized by \emph{Dual Volatility}, enabling rigorous and fair evaluation of existing models. Notably, it is the first benchmark to support both multi-class and multi-label TNC, filling a critical gap in evaluating temporal semantic drift. We conduct a comprehensive evaluation of 17 state-of-the-art methods across Temporal Graph Neural Networks (TGNNs) and Large Language Model (LLM)-based paradigms. The results reveal a clear \emph{capability divide} between the two paradigms: TGNN-based methods excel at structural prediction but fail at semantic tracking, whereas LLM-based predictors show the opposite trend. Through in-depth analysis, we uncover their fundamental limitations and provide insights for developing more comprehensive temporal graph models.
TAHB (Text-Attributed Hypergraph Benchmark) is presented, the first public benchmark integrating hypergraph structures and raw textual attributes, and shows that LLM-enhanced textual semantics improve hypergraph learning performance, while structural and textual information jointly provide the best setting for LLM-based prediction.
D. Y. Kang, Junghyun Kim, Ju-hyun Jeon et al.· 0 citations
With the increasing heterogeneity of social networks and online interaction systems, generalist graph anomaly detection (GAD) has become essential for identifying abnormal and fraudulent behaviors in complex environments. However, most existing GAD approaches rely heavily on domain-specific semantic alignment, which substantially restricts their ability to learn transferable node representations and often leads to poor generalization on unseen graph domains. To address this challenge, we propose HIerarchical Interaction MOdeling for zero-shot generalist GAD (termed HIMO-GAD). HIMO-GAD enables anomaly detection across diverse graph domains without retraining or access to target-domain supervision by modeling the evolutionary trajectories of node representations across hierarchical structural depths, thereby capturing interaction patterns that exhibit strong cross-domain stability. Specifically, HIMO-GAD integrates two core components: (1) a Dynamic Interaction Modeling Module that characterizes cross-layer interaction evolution to extract transferable representations, and (2) an Anomaly-Aware Regulation Mechanism that combines gradient immunity and centralization regularization to suppress overfitting and stabilize cross-domain generalization. Extensive experiments on multiple real-world graph datasets demonstrate that HIMO-GAD consistently outperforms state-of-the-art baselines in strict zero-shot settings, achieving up to a 10% improvement in key evaluation metrics and exhibiting strong generalization across heterogeneous graph domains.
Xiangping Zheng, Xuan Feng, Bo Wu et al.· Proceedings of the 32nd ACM...· 0 citations
A message extrapolation mechanism under soft uncertainty constraints is proposed to obtain the diverse counterfactual message distributions and a novel robust representation learning framework for dynamic graph domain generalization, LEMD is proposed.
Xiaoran Wei, Chen Zhao, Minglai Shao et al.· 0 citations
Large Language Models (LLMs), empowered by autoregressive next-token prediction, have demonstrated strong reasoning capabilities. Extending this paradigm to graph data requires next graph token prediction, yet existing graph tokens struggle to balance two competing requirements: capturing higher-order structural semantics with high information density, and remaining strictly reversible for faithful decoding. Concretely, first-order textual encodings are reversible but long and semantically sparse, while continuous encodings capture high-level semantics but inevitably lose exact topology. To this end, we propose GraphVulcan, a novel framework that enables discrete reversible and semantic-rich tokenization of graphs using a vocabulary of canonical graphlets. Our approach preserves full structural fidelity while enabling LLMs to natively reason over compositional graphlets through graph-level next-token prediction. We propose a three-stage training paradigm: (1) Structural Semantic Pretraining to learn graph token compositionality, (2) Multi-task Fine-tuning on large-scale CoT-augmented reasoning examples, and (3) Reinforcement Learning to explore and refine reasoning paths. Experiments show that GraphVulcan outperforms first-order encoding baselines across 7 graph reasoning tasks and 3 real-world benchmarks while achieving higher computational efficiency. Our work demonstrates that structure-aware discrete tokenization is a feasible way toward general-purpose graph–language models capable of structural reasoning. Codes are available at: https://github.com/alibaba-behavioral-risk-control/GraphVulcan
Zhonghao Wang, Yugang Ji, Zhuonan Zheng et al.· Proceedings of the 32nd ACM...· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
