A Common Language for Dynamic Graphs: Interaction Tokenization Enables Foundation Models

Fuyuan Ma; Yili Wang; Xin He; Xin Zhang; Ying Wang; Xin Wang

cs.LGNov 2, 2026

A Common Language for Dynamic Graphs: Interaction Tokenization Enables Foundation Models

Authors: Fuyuan Ma, Yili Wang, Xin He, Xin Zhang, Ying Wang, Xin Wang

Organizations: School of Artificial Intelligence, Jilin University, Changchun, Jilin, China · School of Computer Science and Technology, Jilin University, Changchun, Jilin, China

Abstract

Dynamic graphs capture evolving interactions in Web systems, including online communication, user engagement, and recommendation. Graph foundation models aim to reuse predictive capabilities across networks, but extending this capability to dynamic graphs faces two challenges. First, node identities are graph-specific, and similar interaction patterns can appear at different temporal and activity scales, making histories difficult to represent consistently across networks. Second, the predictive roles of historical patterns vary across interaction dynamics, limiting transfer from a narrow range of training processes. In this paper, we show that a common interaction language, combined with learning across diverse interaction processes, offers a basis for dynamic graph foundation modeling. We propose CL-DGFM, a Common Language Dynamic Graph Foundation Model for temporal link prediction. Its Dynamic Interaction Tokenizer encodes interaction memory, historical connectivity, and ordered destination transitions into calibrated tabular inputs, making dynamic histories accessible to existing predictors without an additional trainable graph encoder. Its Dynamic Interaction Prior trains a pretrained tabular foundation predictor across diverse synthetic dynamics. Labeled historical context guides prediction, with a Contextual Value Adapter modulating attention values. The model transfers to real graphs without target-graph parameter updates. Experiments on TGB and TGB-Seq demonstrate competitive performance. Further comparisons show that the common language supports basic machine learning and neural models, and enables direct use of frozen tabular foundation models on dynamic graphs.

Figures & tables

Figure 1. Motivation for CL-DGFM. (a) Graph-Domain Foundation Modeling couples encoding with prediction. (b) Foundation Model Adaptation in static settings loses temporal distinctions. (c) An envisioned DGFM uses a common interaction language for shared prediction across dynamic graphs.

Figure 2. Overview of CL-DGFM. (1) The Dynamic Interaction Tokenizer converts real and synthetic histories into a common tabular representation. (2) The Dynamic Interaction Prior generates diverse interaction dynamics for training. (3) Contextual Dynamic Graph Foundation Modeling reuses a pretrained tabular foundation predictor, with a Contextual Value Adapter modulating attention values from labeled history. The model learns on synthetic graphs and ranks candidates on real graphs through historical context, without target-graph parameter updates or an additional trainable graph encoder.

Method	Wiki	Review	Coin	Flight	Comment
CAWN	$71.10_{\pm 0.60}$	$19.30_{\pm 0.10}$	–	–	–
CTAN	$66.80_{\pm 0.70}$	$40.50_{\pm 0.40}$	$74.80_{\pm 0.40}$	–	$67.10_{\pm 6.70}$
DyGFormer	$79.80_{\pm 0.40}$	$22.40_{\pm 1.50}$	$75.20_{\pm 0.40}$	–	$67.00_{\pm 0.10}$
DyRep	$5.00_{\pm 1.70}$	$22.00_{\pm 3.00}$	$45.20_{\pm 4.60}$	$55.60_{\pm 1.40}$	$28.90_{\pm 3.30}$
EdgeBank	$57.10$	$2.50$	$58.00$	$38.70$	$14.90$
GraphMixer	$11.80_{\pm 0.20}$	$\underline{52.10}_{\pm 1.50}$	–	–	–

Table 1. Performance on TGB (

100\times\mathrm{MRR}

; higher is better). Best and second-best results per column are bold red and underlined , respectively.

Method	ML-20M	Taobao	Yelp	GoogleLocal	Flickr	YouTube	Patent	WikiLink
Memory and neighborhood modeling Training per target graph
DyRep	$19.00_{\pm 1.69}$	$40.03_{\pm 2.40}$	$57.69_{\pm 1.05}$	$37.73_{\pm 1.34}$	$38.04_{\pm 4.19}$	$35.12_{\pm 4.13}$	$21.01_{\pm 1.14}$	$42.63_{\pm 1.33}$
JODIE	$21.16_{\pm 0.73}$	$48.36_{\pm 2.18}$	$69.88_{\pm 0.31}$	$41.86_{\pm 1.49}$	$46.21_{\pm 0.83}$	$41.67_{\pm 2.86}$	${\color[rgb]{0.7031,0.1484,0.1953}\mathbf{24.60}}_{\pm 0.38}$	$57.94_{\pm 1.33}$
TGN	$23.99_{\pm 0.20}$	$60.28_{\pm 0.54}$	$69.79_{\pm 0.24}$	$54.13_{\pm 1.97}$	$46.03_{\pm 6.78}$	$55.16_{\pm 5.89}$	$\underline{22.83}_{\pm 2.25}$	$62.94_{\pm 2.16}$
TGAT	$10.47_{\pm 0.20}$	–	–	$19.78_{\pm 0.24}$	$23.53_{\pm 3.35}$	$43.56_{\pm 2.53}$	$8.49_{\pm 0.18}$	–
TCL	$12.04_{\pm 0.02}$	$31.55_{\pm 0.03}$	$24.39_{\pm 0.67}$	$18.30_{\pm 0.02}$	$40.00_{\pm 1.76}$	$50.17_{\pm 1.98}$	$10.60_{\pm 1.75}$	$43.02_{\pm 2.16}$

Table 2. Performance on TGB-Seq (

100\times\mathrm{MRR}

; higher is better). Baselines train per target graph; CL-DGFM uses a shared model with labeled context and fixed parameters. Each column’s best result is bold red and its second best is underlined .

Predictor	Wiki	Google Local	Flickr	YouTube
Machine learning Training per graph
Logistic regression	$79.73$	$54.31$	$70.07$	$69.32$
Random forest	$79.25$	$56.10$	$71.08$	$69.51$
CatBoost	$80.65$	$56.13$	$72.20$	$69.95$
LightGBM	$81.21$	$54.12$	$71.94$	$66.63$
Deep learning Training per graph

Table 3. Predictor comparison using the same tokenizer (

100\times\mathrm{MRR}

; higher is better). Column-wise best results are red and bold ; second-best results are underlined .

Figure 3. Representation comparison with a frozen backbone. Error bars show standard deviations over three paired context draws; dashed lines mark the full tokenizer.

Figure 4. MRR changes from ablating (a) tokenizer components (RQ2) and (b) prior mechanisms (RQ3).

Figure 5. Training on more distinct synthetic graphs versus repeatedly using fewer graphs, with the same number of training updates and graph samples. Shades indicate graph-pool sizes; endpoints report MRR.

Figure 6. Effect of context size on prediction (

100\times

MRR).

K

counts positive context events, each paired with one negative. Dashed lines mark the main setting

K=512

; vertical-axis ranges differ across datasets.

Dataset	Domain (type)	Nodes	Events	Test	Neg.
Synthetic training data
Generated streams	Synthetic interactions (B/H)	$\approx 1.30$ M	$\approx 28.50$ M(training data)	–	–
TGB evaluation datasets
Wiki	Page edits (B)	9,227	157,474	23,621	999
Review	Product reviews (B)	352,637	4,873,540	728,919	100
Coin	Transactions (H)	638,486	22,809,486	3,421,411	20

Table 4. Statistics of synthetic training data and real evaluation datasets. B/H denote bipartite/homogeneous graphs. Events counts all observed interactions; Test counts evaluation queries. Neg. denotes negative destinations per test query, excluding the positive destination.

Appendix figures & tables2 assets

Supplementary material from the paper’s appendix.

Appendix

No.	Features and settings
Interaction memory: $\mathbf{z}^{\mathrm{mem}}\in\mathbb{R}^{18}$
1–4	$\ell_{+}(M_{E_{uv}}(h)),\ h\in\mathcal{B}$ , using all past $u\to v$ events
5–8	$\ell_{+}(M_{I_{v}}(h)),\ h\in\mathcal{B}$
9–11	$\ell_{+}(C_{uv}),\ \ell_{+}(d_{u}),\ \ell_{+}(d_{v})$
12–14	$R_{E_{uv}},\ R_{I_{u}},\ R_{I_{v}}$
15–17	$\ell_{+}(C_{vu}),\ \mathbf{1}\{C_{uv}>0\},\ \mathbf{1}\{d_{v}=0\}$

Appendix

Table 5. The 31-dimensional history descriptor.

\mathcal{B}=s\{0.03,0.3,3,30\}

is ordered increasingly. Ratios precede log compression, and

h=\infty

removes decay.

Setting	Value
Stream length	$\{3{,}000,6{,}000,12{,}000\}$ events
Positive context events	$\{64,128,256,512\}$ with probabilities $(0.3,0.3,0.3,0.1)$
Context / query interval	$30$ – $65\%$ / $65$ – $95\%$
Queries / candidates	$8$ / $\{32,64,128\}$ , capped by the destination pool
Optimizer / learning rates	AdamW, $10^{-5}$ for the predictor and $10^{-3}$ for the adapter
Updates / tasks per update	$1{,}024$ / $4$ , with $4{,}096$ updates for the diversity study

Appendix

Table 6. Task construction and training settings. Intervals refer to positions in the generated stream.

Graph Foundation Models

Graph Foundation Models are pretrained graph-learning systems designed to transfer across datasets, tasks, and domains with different structures and features.

See all 101 papers

Oct 14, 2026

Dynamic Graph World Modelling: Learning Future Dynamics from Temporal Context

Dynamic graph learning models how entities and their relations evolve from historical observations to predict future graph states. Existing methods do not fully address dynamic graph world modelling, which requires learning transitions from temporal context to future latent graph states. This is difficult for three reasons: nodes and relations rely on different historical evidence, structural evolution is path-dependent, and future snapshots provide sparse and realization-specific targets. We propose Dynamic Graph World Modelling (DyG-WM), the first framework to jointly predict future node and relation representations from temporal context by answering three questions: what has evolved, how it evolved, and what will evolve next. To capture what has evolved, a historical node-relation context encoder summarizes node-neighborhood trajectories and pair-specific histories. To characterize how it evolved, a node-relation path-signature encoder captures ordered structural changes. To predict what will evolve next, a dual-level joint-embedding objective predicts target-snapshot node and relation representations without reconstructing the raw graph. Experiments on 12 link-prediction and two node-classification datasets demonstrate the effectiveness of DyG-WM across both tasks.

Bo Li, Yue Yang, Yujin Huang +6

School of Information and Communication Technology, Griffith University, Australia · Maincode, Australia · School of Computing and Information Systems, University of Melbourne, Australia +2

Dynamic Graph Learning World Model Learning Predictive Representation Learning Temporal Link Prediction Joint-Embedding Predictive Architecture

Ask PDF

May 13, 2026

Decoupled and Divergence-Conditioned Prompt for Multi-domain Dynamic Graph Foundation Models

Dynamic graphs are ubiquitous in real-world systems, and building generalizable dynamic Graph Foundation Models has become a frontier in graph learning. However, dynamic graphs from different domains pose fundamental challenges to unified modeling, as their semantic and temporal patterns are inherently inconsistent, making the multi-domain pre-training difficult. Consequently, the widely used "pretrain-then-finetune" paradigm often suffers from severe negative knowledge transfer. To the best of our knowledge, there exists no multi-domain dynamic GFM. In this work, we propose DyGFM, a Dynamic Graph Foundation Model over multiple domains based on decoupled and divergence-conditioned prompting. To disentangle transferable semantics from the domain-specific dynamics, we introduce a dual-branch pre-training strategy with semantic-temporal decoupling. To alleviate negative transfer during domain adaptation, we further develop a cross-domain routing mechanism with divergence-aware expert selection. To enable efficient downstream fine-tuning, we design a divergence-conditioned prompt generator that injects lightweight, learnable graph prompts tailored to semantic and temporal traits. Extensive experiments on continuous dynamic graph benchmarks demonstrate that DyGFM consistently outperforms 12 state-of-the-art baselines on both node classification and link prediction tasks, achieving superior effectiveness and efficiency.

Haonan Yuan, Qingyun Sun, Junhua Shi +3

School of Computer Science and Engineering, Beihang University, Beijing 100191, China · Key Lab of Education Blockchain and Intelligent Technology, Ministry of Education, Guangxi Normal University, Guilin 541004, China · Department of Computer Science, University of Illinois at Chicago, Chicago 60607, USA

Domain-Adaptive Pretraining Prompt Tuning Cross-Domain Transfer Learning Dynamic Graph Learning Graph Foundation Models

Ask PDF

Mar 9, 2026

Graph2Video: Leveraging Video Models to Model Dynamic Graph Evolution

Dynamic graphs are common in real-world systems such as social media, recommender systems, and traffic networks. Existing dynamic graph models for link prediction often fall short in capturing the complexity of temporal evolution. They tend to overlook fine-grained variations in temporal interaction order, struggle with dependencies that span long time horizons, and offer limited capability to model pair-specific relational dynamics. To address these challenges, we propose Graph2Video, a video-inspired framework that views the temporal neighborhood of a target link as a sequence of "graph frames". By stacking temporally ordered subgraph frames into a "graph video", Graph2Video leverages the inductive biases of video foundation models to capture both fine-grained local variations and long-range temporal dynamics. It generates a link-level embedding that serves as a lightweight and plug-and-play link-centric memory unit. This embedding integrates seamlessly into existing dynamic graph encoders, effectively addressing the limitations of prior approaches. Extensive experiments on benchmark datasets show that Graph2Video outperforms state-of-the-art baselines on the link prediction task in most cases. The results highlight the potential of borrowing spatio-temporal modeling techniques from computer vision as a promising and effective approach for advancing dynamic graph learning.

Hua Liu, Yanbin Wei, Fei Xing +3

Southern University of Science and Technology · The Hong Kong University of Science and Technology · City University of Hong Kong +2

Dynamic Graph Learning Temporal Link Prediction Graph Embedding

Ask PDF

Abstract

Graph Foundation Models

Explore similar work

Dynamic Graph World Modelling: Learning Future Dynamics from Temporal Context

Decoupled and Divergence-Conditioned Prompt for Multi-domain Dynamic Graph Foundation Models

Graph2Video: Leveraging Video Models to Model Dynamic Graph Evolution