Pith. sign in

REVIEW 3 major objections 5 minor 56 references

Beyond Feature and Structure Alignment: Learning Transferable Propagation Knowledge for Graph Foundation Models

T0 review · 3 major / 5 minor · reviewed 2026-08-03 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read Propagation relationships between edges and feature dimensions are transferable knowledge units that let a graph model generalize without fine-tuning.

desk verdict A clean idea with solid node-level results, but the transfer story has an unaddressed SVD alignment hole and the graph-classification table doesn't include ProGFM. read the letter →

arxiv 2607.28980 v1 pith:S6JX4SJN submitted 2026-07-31 cs.LG

Yi Wang , Jitao Zhao , Di Jin , Dongxiao He This is my paper
classification cs.LG
keywords graphfoundationmodelsknowledgetransferpropagationrelationshipsprototypebankmessagepassingcross-domaingeneralizationself-supervisedpre-trainingrelativefeaturedifference
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper argues that the missing transferable unit in graph foundation models is neither a token nor a patch but the propagation relationship between an edge and a feature dimension: how similar or dissimilar the two connected nodes are along that dimension. The authors propose ProGFM, which encodes each edge-dimension pair as a scalar derived from the relative feature difference, clusters these scalars from source graphs into a prototype bank, and attaches a learnable propagation strength to each prototype. A propagation-aware message passing rule then modulates each feature dimension of each neighbor's message by the matched strength. The paper's claim is that this bank of propagation patterns, once learned on source domains, transfers directly to unseen graph domains even when feature semantics and structures differ, because the patterns depend on relative differences rather than absolute meanings.

What carries the argument

The central object is the propagation relationship scalar s_{ij,k}=1−r_{ij,k}, where r_{ij,k} is the relative feature difference between two connected nodes on feature dimension k. This scalar carries whether an edge tends to preserve or suppress information along that dimension: values near 1 mean connected nodes are similar on that dimension (homophily-like), values near 0 mean they are far apart. The prototype bank, built by K-means clustering of these scalars across source graphs, is a set of pairs (p_c, alpha_c) where p_c is a representative propagation relationship and alpha_c a learned propagation strength. The propagation-aware message passing in Eq. 18 does the main work: for each e

What would settle it

Synthesize two graphs with identical edge-wise relative feature differences but opposite label-relationship patterns: in one, connected nodes with similar feature values share labels; in the other, connected nodes with dissimilar values share labels. Pre-train ProGFM on the first, freeze it, and evaluate on the second. If the model transfers without catastrophic loss, the relative-difference assumption survives; if accuracy collapses to chance, the propagation-relationship claim is refuted.

Watch

Extended reading notes

Core claim

The central claim is that propagation relationships — defined as s_{ij,k} = 1 − r_{ij,k}, where r is the normalized absolute feature difference between connected nodes on dimension k — can serve as domain-agnostic transferable knowledge units. ProGFM first projects all graphs to a shared feature dimensionality via SVD, then collects s-values across all edges and dimensions from pre-training graphs, clusters them with K-means into a propagation relationship prototype bank, and associates each prototype with a learnable scalar propagation strength. At each layer, the model recomputes the s-values in the current message space, matches each to its nearest prototype, retrieves the corresponding s

Load-bearing premise

The claim stands or falls on the assumption that equal relative feature differences between connected nodes always entail equal propagation behavior, so a prototype bank learned from source-domain differences stays valid on target domains after SVD alignment and across completely different feature semantics.

Editorial extensions

If this is right

  • A frozen pre-trained model can be applied to an unseen domain with only labeled class prototypes, no gradient updates.
  • Propagation knowledge transfers across task granularities, from node classification pre-training to graph classification.
  • Each edge can multiply each feature dimension of its neighbor's message by a learned strength, enabling fine-grained adaptive propagation that fixed rules cannot express.
  • Only dimensionality alignment, not semantic alignment, is needed across domains, since propagation relationships are computed from relative differences.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • A natural extension would be to replace the hard nearest-prototype lookup with a continuous learned function from s-values to strengths, eliminating potential discontinuities at prototype boundaries.
  • The learned strengths could be interpreted as a data-driven notion of per-dimension homophily across domains; inspecting them might reveal which feature dimensions are consistently smoothing vs. sharpening across graphs.
  • The same logic suggests other relative measures — rank differences, normalized dot products, or angle-based similarities — could serve as alternative transferable units; comparing them would map the boundary of the claim.
  • The paper's zero-tuning protocol is strict; a lighter-weight consequence is that even partially updating only the prototype strengths could close most of the gap to full fine-tuning, which the ProGFM-tuning variant already hints at.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 5 minor

Summary. The paper proposes ProGFM, a graph foundation model that treats the relationship between each edge and each feature dimension as a transferable knowledge unit. It first projects node features of each graph into a common dimensionality via SVD, then computes a scalar propagation relationship s_ij,k = 1 - r_ij,k from the normalized feature difference across an edge. K-means clustering of these scalars across source graphs forms a prototype bank B = {(p_c, alpha_c)}. During message passing, each edge receives a propagation-strength vector by quantizing its current-layer propagation relationships to the nearest prototype; the message is z_ij \odot W h_j. ProGFM is pre-trained with the SGRL objective on multiple source graphs and then transferred to unseen target graphs with all parameters frozen (zero-tuning), with predictions made by prototype-based classification using labeled target nodes. Experiments cover one-shot/few-shot node classification, subgraph classification, and graph classification.

Significance. If the proposed transfer mechanism is sound, treating propagation relationships (quantified by relative feature differences) as knowledge units is a genuinely different approach from feature/structure alignment and could be a useful building block for graph foundation models. The empirical study is broad: leave-one-graph-out node classification and subgraph classification with 500 few-shot tasks per target, plus graph classification across social and protein domains, with a range of GNN, self-supervised, and GFM baselines. The consistent improvements on most node/subgraph settings are encouraging. However, the paper does not yet establish the central claim because (i) per-graph SVD leaves the coordinate systems unaligned, so the 'domain-agnostic' quantities are not comparable across domains; (ii) the graph-classification table omits the proposed method; and (iii) the default K is selected on target data, which conflicts with the zero-tuning claim. These issues are fixable but are not merely editorial.

major comments (3)
  1. [§3.2–§3.4, Eqs. (3), (10), (14)] The transfer mechanism is built on a coordinate system that is not shared across domains. Eq. (3) aligns feature dimensionality with a per-graph SVD. SVD bases are dataset-specific and only defined up to sign (and arbitrary rotation in degenerate subspaces), so coordinate k in one graph is not the same latent direction as coordinate k in another. Nevertheless, Eq. (10) applies a shared W^(l) to these representations and Eq. (14) matches the scalar s^(l)_ij,k to a global prototype p_c. Thus both the prototype matching and the shared linear transformation presuppose that the coordinate index k is a universal slot, which the paper explicitly declines to ensure (Sec. 3.1). If the per-graph projections are not aligned, the values s_ij,k from different domains are not comparable scalars and the 'transferable propagation knowledge' reduces to a scalar quantization of domain-dependent coordinate
  2. [Table 4 / §4.6] The graph-classification experiment is not reported as claimed. Table 4 contains no row labelled ProGFM; the row labelled 'FLT' is not defined in §4.2 or anywhere else, and §4.6 states that 'ProGFM achieves the best performance on all evaluated graph classification datasets.' As written, the reader cannot verify the central graph-classification claim. The table needs a correctly labelled ProGFM row (or the names must be fixed) and the accompanying text must refer to the row actually present.
  3. [§4.8 / Fig. 3] The zero-tuning claim is weakened by the selection of K. The paper says 'Based on the overall performance across different datasets, we set K=100 as the default configuration' (Sec. 4.8), where the 'different datasets' appear to be the target datasets used in the sensitivity analysis. Under the zero-tuning protocol of Eq. (21), target-domain data should not be used to choose model configuration. Please state how K is selected using only source-domain information, or explicitly report K as a fixed architectural choice with sensitivity analysis as a post-hoc diagnostic; otherwise the comparison is not a fair zero-tuning evaluation.
minor comments (5)
  1. [Fig. 2 / Fig. 3] The ablation and sensitivity figures do not include error bars or significance statements. Given that Tables 1–2 report standard deviations over 500 tasks, the figures should include them as well.
  2. [Eq. (15)] Notation: Eq. (8)/(9) denote learnable strengths as alpha_c, while Eq. (15) writes B_c=(p_c,a_c). Use a consistent symbol.
  3. [Eq. (17)] Typo: 'where where z^(l)_ij denotes' should be 'where z^(l)_ij denotes'.
  4. [Table 1] The TIG row has a formatting issue: '45.50±8.7651.79±9.98' lacks a separator between the CiteSeer and PubMed entries.
  5. [§4.7] The 'w/o Learnable Strength' variant initializes fixed propagation strengths from N(1,1); since this distribution puts probability mass on negative values, please clarify whether negative propagation strengths are intended or should be restricted to positive values.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: propagation relationships are data-derived features, alpha are ordinary learned parameters, and transfer is validated on held-out standard benchmarks; same-group citations are not load-bearing.

full rationale

No load-bearing circular step is present. The propagation relationship s_ij,k is defined by an explicit feature-difference formula (Eqs. 1-4), the prototype bank is built by K-means over these computed values (Eqs. 5-7), and the propagation strengths alpha are ordinary learnable parameters optimized under a self-supervised loss (Eq. 20). No 'prediction' in the paper is obtained by renaming a fitted value or by a definitional identity. The cross-domain transferability claim is an empirical hypothesis validated on standard held-out datasets (Cora/CiteSeer/PubMed/Photo/Computers/CS and IMDB-BINARY/COLLAB/PROTEINS/DD) under leave-one-graph-out and zero-tuning protocols, so it has independent evidential content. The paper does cite prior work by the same group ([7], [27], [35], [48], [50]) for the SSL objective and baselines, but those citations are not used to derive the central transfer claim and would not by themselves force the results. The per-domain SVD coordinate-alignment concern raised in the skeptic note is a correctness/robustness risk, not a circularity, because it attacks an assumption rather than exhibiting a reduction of output to input by construction.

Assumptions & free parameters 4 free parameters · 5 assumptions · 2 invented entities

The central contribution is an architecture whose prototypes and strengths are fitted to source data; the transferability of those prototypes is the key unproven assumption. No external physical or formal benchmark independently validates the proposed knowledge units.

free parameters (4)
  • alpha_c (propagation strength per prototype) = learned during pretraining; values not reported
    Each prototype p_c has a learnable scalar alpha_c (Eqs. 8-9) that modulates aggregation; these are optimized with the pretraining objective (Eq. 20).
  • K (number of prototypes) = 100
    Chosen via sensitivity analysis on target datasets (Section 4.8); performance peaks at different K per dataset, but K=100 is used as the default.
  • d (unified feature dimensionality) = 16 for node/subgraph, 128 for graph classification
    SVD output dimension chosen per scenario (Table 6); it determines the granularity of per-dimension propagation.
  • Layer widths (Dim #1, Dim #2) = 256/128 or 512/512 depending on scenario
    Architecture hyperparameters (Table 6).
assumptions (5)
  • domain assumption Propagation relationships between edges and feature dimensions are transferable across graph domains
    Core premise stated in Sections 1 and 3.3; no theoretical justification, supported only by the paper's experiments.
  • domain assumption Similar relative feature differences imply similar propagation patterns
    Assumed in Section 3.3 to justify using s=1-r as the knowledge unit; not proven.
  • domain assumption SVD dimensionality alignment does not destroy transferable relative differences
    ProGFM aligns features via SVD (Section 3.2); the relative differences of the projected features are assumed to retain cross-domain transferability.
  • standard math K-means clustering yields meaningful propagation prototypes
    K-means is standard, but the choice of K and the feature space is heuristic (Section 3.3).
  • domain assumption SGRL self-supervised objective provides a useful pretraining signal
    Adopted from [7] (Section 3.5); its effectiveness for this architecture is assumed.
invented entities (2)
  • Propagation relationship prototype bank B
    purpose: Stores K prototypes (p_c) with learnable strengths alpha_c; used to assign propagation strengths to edges and feature dimensions in unseen domains
    An architectural construct; its transferability is asserted and tested only within this paper's benchmarks.
  • Propagation relationship knowledge units (s_ij,k)
    purpose: Conceptual reification of relative feature differences as transferable knowledge
    No independent falsifiable handle outside the proposed model; the concept is defined by the method itself.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Beyond Feature and Structure Alignment: Learning Transferable Propagation Knowledge for Graph Foundation Models." pith.science (2026). https://pith.science/paper/S6JX4SJN

@misc{pith2026260728980,
  author       = {Pith},
  title        = {Pith review of: Beyond Feature and Structure Alignment: Learning Transferable Propagation Knowledge for Graph Foundation Models},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/S6JX4SJN}},
  note         = {Machine review of arXiv:2607.28980}
}
read the original abstract

Graph Foundation Models (GFMs) have recently emerged as a promising paradigm for enabling knowledge transfer across diverse domains. Unlike traditional graph learning methods that are typically designed for in-domain settings, GFMs aim to learn transferable knowledge that can generalize to unseen graph domains. However, unlike language or visual data, graphs lack intrinsic and unified representation units, such as tokens in language and patches in vision, making it challenging to identify transferable knowledge units for building graph foundation models. Existing graph foundation models mainly focus on mitigating domain discrepancies through feature alignment and structure alignment, while overlooking the exploration of transferable knowledge units underlying graph data. Moreover, these methods generally rely on fixed propagation mechanisms during message passing, overlooking the heterogeneity in propagation patterns, as different edges may exhibit distinct propagation patterns for different feature dimensions. To address these limitations, we propose a Propagation-aware Graph Foundation Model (ProGFM), which regards the propagation relationships between edges and feature dimensions as transferable knowledge units. Through a propagation relationship prototype bank, ProGFM learns cross-domain transferable propagation knowledge, enabling adaptive information aggregation in unseen graph domains. Extensive experiments across various cross-domain transfer scenarios demonstrate that ProGFM possesses strong cross-domain knowledge transfer capability and exhibits superior generalization performance compared with existing methods.

Figures

Figures reproduced from arXiv: 2607.28980 by the authors.

Figure 1
Figure 1. The overview of ProGFM. Each propagation relationship prototype is further associated with a learnable scalar parameter that represents its corresponding prop￾agation strength. Finally, we introduce a propagation-aware mes￾sage passing mechanism. Based on the propagation relationships between each edge and each feature dimension, each edge is associ￾ated with a propagation strength vector, which is used to modulate … view at source ↗
Figure 2
Figure 2. Ablation study results of ProGFM. The effectiveness [PITH_FULL_IMAGE:figures/full_fig_p008_2.png] view at source ↗
Figure 3
Figure 3. Sensitivity analysis of the number of propagation [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

56 extracted references · 4 canonical work pages

  1. [42]

    Liang Yang, Mengzhe Li, Liyang Liu, bingxin niu, Chuan Wang, Xiaochun Cao, and Yuanfang Guo. 2021. Diverse Message Passing for Attribute with Heterophily. InAdvances in Neural Information Processing Systems, M. Ranzato, A. Beygelz- imer, Y. Dauphin, P.S. Liang, and J. Wortman Vaughan (Eds.), Vol. 34. Curran Associates, Inc., 4751–4763. https://proceedings...

    2021

  2. [1]

    Hervé Abdi and Lynne J Williams. 2010. Principal component analysis.Wiley interdisciplinary reviews: computational statistics2, 4 (2010), 433–459

    2010

  3. [2]

    Borgwardt, Cheng Soon Ong, Stefan Schönauer, S

    Karsten M. Borgwardt, Cheng Soon Ong, Stefan Schönauer, S. V. N. Vishwanathan, Alexander J. Smola, and Hans-Peter Kriegel. 2005. Protein function prediction via graph kernels. InProceedings Thirteenth International Conference on Intelligent Systems for Molecular Biology 2005, Detroit, MI, USA, 25-29 June 2005. 47–56. doi:10.1093/BIOINFORMATICS/BTI1007

    doi:10.1093/bioinformatics/bti1007 2005

  4. [3]

    Paul D Dobson and Andrew J Doig. 2003. Distinguishing enzyme structures from non-enzymes without alignments.Journal of molecular biology330, 4 (2003), 771–783

    2003

  5. [4]

    Moshe Eliasof, Krishna Sri Ipsit Mantri, Beatrice Bevilacqua, Bruno Ribeiro, and Carola-Bibiane Schönlieb. 2026. Bridging Input Feature Spaces Towards Graph Foundation Models.CoRRabs/2605.04834 (2026). arXiv:2605.04834 doi:10.48550/ ARXIV.2605.04834

    work page Pith review arXiv doi:10.48550/arxiv.2605.04834 2026

  6. [5]

    Wenqi Fan, Yao Ma, Qing Li, Yuan He, Yihong Eric Zhao, Jiliang Tang, and Dawei Yin. 2019. Graph Neural Networks for Social Recommendation. InThe World Wide Web Conference, WWW 2019, San Francisco, CA, USA, May 13-17, 2019, Ling Liu, Ryen W. White, Amin Mantrach, Fabrizio Silvestri, Julian J. McAuley, Ricardo Baeza-Yates, and Leila Zia (Eds.). ACM, 417–426...

    arXiv 2019

  7. [6]

    Chen Gao, Yu Zheng, Nian Li, Yinfeng Li, Yingrong Qin, Jinghua Piao, Yuhan Quan, Jianxin Chang, Depeng Jin, Xiangnan He, and Yong Li. 2023. A Survey of Graph Neural Networks for Recommender Systems: Challenges, Methods, and Directions.Trans. Recomm. Syst.1, 1 (2023), 1–51. doi:10.1145/3568022

    work page doi:10.1145/3568022 2023

  8. [7]

    Dongxiao He, Lianze Shan, Jitao Zhao, Hengrui Zhang, Zhen Wang, and Weixiong Zhang. 2024. Exploitation of a Latent Mechanism in Graph Contrastive Learning: Representation Scattering. InThe Thirty-eighth Annual Conference on Neural Information Processing Systems

    2024

Show all 56 references
  1. [8]

    Yufei He, Yuan Sui, Xiaoxin He, and Bryan Hooi. 2025. UniGraph: Learning a Uni- fied Cross-Domain Foundation Model for Text-Attributed Graphs. InProceedings of the 31st ACM SIGKDD Conference on Knowledge Discovery and Data Mining, V.1, KDD 2025, Toronto, ON, Canada, August 3-7...

    2025

  2. [9]

    Zhenyu Hou, Xiao Liu, Yukuo Cen, Yuxiao Dong, Hongxia Yang, Chunjie Wang, and Jie Tang. 2022. GraphMAE: Self-Supervised Masked Graph Autoencoders. InKDD ’22: The 28th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, Washington, DC, USA, August 14 - 18, 2022. ACM, ...

    2022

  3. [10]

    Lokesh Jain, Rahul Katarya, and Shelly Sachdeva. 2023. Opinion Leaders for Information Diffusion Using Graph Neural Network in Online Social Networks. ACM Trans. Web17, 2, Article 13 (April 2023), 37 pages. doi:10.1145/3580516

    2023 doi

  4. [11]

    Mohsen Jamali and Martin Ester. 2010. A matrix factorization technique with trust propagation for recommendation in social networks. InProceedings of the 2010 ACM Conference on Recommender Systems, RecSys 2010, Barcelona, Spain, September 26-30, 2010, Xavier Amatriain, Marc To...

    2010

  5. [12]

    Weiwei Jiang and Jiayun Luo. 2022. Graph neural network for traffic forecasting: A survey.Expert Systems with Applications207 (2022), 117921. doi:10.1016/j.eswa. 2022.117921

    2022

  6. [13]

    Kipf and Max Welling

    Thomas N. Kipf and Max Welling. 2017. Semi-Supervised Classification with Graph Convolutional Networks. In5th International Conference on Learning Representations, ICLR 2017, Toulon, France, April 24-26, 2017, Conference Track Proceedings. OpenReview.net. https://openreview.ne...

    2017

  7. [14]

    Lecheng Kong, Jiarui Feng, Hao Liu, Chengsong Huang, Jiaxin Huang, Yixin Chen, and Muhan Zhang. 2025. GOFA: A Generative One-For-All Model for Joint Graph Language Modeling. InThe Thirteenth International Conference on Learning Representations, ICLR 2025, Singapore, April 24-2...

    2025

  8. [15]

    Haoyang Li, Haibo Chen, Xin Wang, and Wenwu Zhu. 2026. Out-of-Distribution Generalization in Graph Foundation Models.CoRRabs/2601.21067 (2026). arXiv:2601.21067 doi:10.48550/ARXIV.2601.21067

    2026 doi

  9. [16]

    Hao Liu, Jiarui Feng, Lecheng Kong, Ningyue Liang, Dacheng Tao, Yixin Chen, and Muhan Zhang. 2023. One for All: Towards Training One Graph Model for All Classification Tasks.CoRRabs/2310.00149 (2023). arXiv:2310.00149 doi:10.48550/ARXIV.2310.00149

    2023 doi arXiv

  10. [17]

    Jingzhe Liu, Haitao Mao, Zhikai Chen, Wenqi Fan, Mingxuan Ju, Tong Zhao, Neil Shah, and Jiliang Tang. 2024. One Model for One Graph: A New Perspec- tive for Pretraining with Cross-domain Graphs.CoRRabs/2412.00315 (2024). arXiv:2412.00315 doi:10.48550/ARXIV.2412.00315

    2024 doi arXiv

  11. [18]

    Jiawei Liu, Cheng Yang, Zhiyuan Lu, Junze Chen, Yibo Li, Mengmei Zhang, Ting Bai, Yuan Fang, Lichao Sun, Philip S Yu, et al. 2023. Towards graph foundation models: A survey and beyond.arXiv preprint arXiv:2310.11829(2023)

    2023 arXiv

  12. [19]

    Lei Liu, Xingyu Xia, Qianqian Xie, Ben Liu, Wenjie Xu, and Min Peng. 2025. Enhanced Expert Merging for Mixture-of-Experts in Graph Foundation Models. InAdvances in Neural Information Processing Systems, D. Belgrave, C. Zhang, H. Lin, R. Pascanu, P. Koniusz, M. Ghassemi, and N....

    2025

  13. [20]

    Tong Liu and Hadi Meidani. 2024. End-to-end heterogeneous graph neural networks for traffic assignment.Transportation Research Part C: Emerging Tech- nologies165 (2024), 104695. doi:10.1016/j.trc.2024.104695

    2024

  14. [21]

    Stuart P. Lloyd. 1982. Least squares quantization in PCM.IEEE Trans. Inf. Theory 28, 2 (1982), 129–136. doi:10.1109/TIT.1982.1056489

    1982

  15. [22]

    Haitao Mao, Zhikai Chen, Wenzhuo Tang, Jianan Zhao, Yao Ma, Tong Zhao, Neil Shah, Mikhail Galkin, and Jiliang Tang. 2024. Position: Graph Foundation Models Are Already Here. InForty-first International Conference on Machine Learning, ICML 2024, Vienna, Austria, July 21-27, 202...

    2024

  16. [23]

    McAuley, Christopher Targett, Qinfeng Shi, and Anton van den Hengel

    Julian J. McAuley, Christopher Targett, Qinfeng Shi, and Anton van den Hengel

  17. [24]

    Kriege, Franka Bause, Kristian Kersting, Petra Mutzel, and Marion Neumann

    Christopher Morris, Nils M. Kriege, Franka Bause, Kristian Kersting, Petra Mutzel, and Marion Neumann. 2020. TUDataset: A collection of benchmark datasets for learning with graphs.CoRRabs/2007.08663 (2020). arXiv:2007.08663 https: //arxiv.org/abs/2007.08663

    2020 arXiv

  18. [25]

    Prithviraj Sen, Galileo Namata, Mustafa Bilgic, Lise Getoor, Brian Gallagher, and Tina Eliassi-Rad. 2008. Collective Classification in Network Data.AI Mag.29, 3 (2008), 93–106. doi:10.1609/AIMAG.V29I3.2157

    2008 doi

  19. [26]

    Lianze Shan, Jitao Zhao, Dongxiao He, Yongqi Huang, Zhiyong Feng, and Weix- iong Zhang. 2026. MUG: Meta-path-aware Universal Heterogeneous Graph Pre-Training. InFortieth AAAI Conference on Artificial Intelligence, Thirty-Eighth Conference on Innovative Applications of Artifici...

    2026

  20. [27]

    Lianze Shan, Jitao Zhao, Dongxiao He, Siqi Liu, Jiaxu Cui, and Weixiong Zhang

  21. [28]

    Arnab Sinha, Zhihong Shen, Yang Song, Hao Ma, Darrin Eide, Bo-June Paul Hsu, and Kuansan Wang. 2015. An Overview of Microsoft Academic Service (MAS) and Applications. InProceedings of the 24th International Conference on World Wide Web Companion, WWW 2015, Florence, Italy, May...

    2015

  22. [29]

    Li Sun, Zhenhao Huang, Suyang Zhou, Qiqi Wan, Hao Peng, and Philip S. Yu

  23. [30]

    Shantanu Thakoor, Corentin Tallec, Mohammad Gheshlaghi Azar, Rémi Munos, Petar Veličković, and Michal Valko. 2021. Bootstrapped representation learning on graphs. InICLR 2021 Workshop on Geometrical and Topological Representation Learning

    2021

  24. [31]

    Petar Velickovic, Guillem Cucurull, Arantxa Casanova, Adriana Romero, Pietro Liò, and Yoshua Bengio. 2018. Graph Attention Networks. In6th International Conference on Learning Representations, ICLR 2018, Vancouver, BC, Canada, April 30 - May 3, 2018, Conference Track Proceedin...

    2018

  25. [32]

    Hamilton, Pietro Liò, Yoshua Bengio, and R

    Petar Velickovic, William Fedus, William L. Hamilton, Pietro Liò, Yoshua Bengio, and R. Devon Hjelm. 2019. Deep Graph Infomax. In7th International Conference on Learning Representations, ICLR 2019, New Orleans, LA, USA, May 6-9, 2019. OpenReview.net. https://openreview.net/for...

    2019

  26. [33]

    Kai Wang, Siqiang Luo, Caihua Shan, and Yifei Shen. 2025. Towards Graph Foundation Models: Training on Knowledge Graphs Enables Transferability to General Graphs. InAdvances in Neural Information Processing Systems 38: Annual Conference on Neural Information Processing Systems...

    2025

  27. [34]

    Shuo Wang, Bokui Wang, Zhixiang Shen, Boyan Deng, and Zhao Kang. 2025. Multi-Domain Graph Foundation Models: Robust Knowledge Transfer via Topol- ogy Alignment.CoRRabs/2502.02017 (2025). arXiv:2502.02017 doi:10.48550/ ARXIV.2502.02017

    2025 doi arXiv

  28. [35]

    Yi Wang, Jitao Zhao, Dongxiao He, Jia Li, Yuxiao Huang, and Zhiyong Feng. 2026. Topology-Aware Feature Sorting Enables Universal Modeling on Homophilic and , , Yi Wang et al. Heterophilic Graphs. InProceedings of the ACM Web Conference 2026, WWW 2026, Dubai, United Arab Emirat...

    2026

  29. [36]

    Chawla, Chuxu Zhang, and Yanfang Ye

    Zehong Wang, Zheyuan Liu, Tianyi Ma, Jiazheng Li, Zheyuan Zhang, Xingbo Fu, Yiyang Li, Zhengqing Yuan, Wei Song, Yijun Ma, Qingkai Zeng, Xiusi Chen, Jianan Zhao, Jundong Li, Meng Jiang, Pietro Lio, Nitesh V. Chawla, Chuxu Zhang, and Yanfang Ye. 2025. Graph Foundation Models: A...

    2025 doi arXiv

  30. [37]

    Chawla, Chuxu Zhang, and Yan- fang Ye

    Zehong Wang, Zheyuan Zhang, Nitesh V. Chawla, Chuxu Zhang, and Yan- fang Ye. 2024. GFT: Graph Foundation Model with Transferable Tree Vocabulary. InAdvances in Neural Information Processing Systems 37: An- nual Conference on Neural Information Processing Systems 2024, NeurIPS ...

    2024

  31. [38]

    Shiwen Wu, Fei Sun, Wentao Zhang, Xu Xie, and Bin Cui. 2023. Graph Neural Networks in Recommender Systems: A Survey.ACM Comput. Surv.55, 5 (2023), 97:1–97:37. doi:10.1145/3535101

    2023 doi

  32. [39]

    Zonghan Wu, Shirui Pan, Fengwen Chen, Guodong Long, Chengqi Zhang, and S Yu Philip. 2021. A comprehensive survey on graph neural networks.IEEE Trans. Neural Networks Learn. Syst.32, 1 (2021), 4–24. https://doi.org/10.1109/ TNNLS.2020.2978386

    2021

  33. [40]

    Lianghao Xia and Chao Huang. 2024. Anygraph: Graph foundation model in the wild. (2024)

    2024

  34. [41]

    Pinar Yanardag and S. V. N. Vishwanathan. 2015. Deep Graph Kernels. InProceed- ings of the 21th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, Sydney, NSW, Australia, August 10-13, 2015, Longbing Cao, Chengqi Zhang, Thorsten Joachims, Geoffrey I. W...

    2015

  35. [43]

    Cohen, and Ruslan Salakhutdinov

    Zhilin Yang, William W. Cohen, and Ruslan Salakhutdinov. 2016. Revisiting Semi-Supervised Learning with Graph Embeddings. InProceedings of the 33nd International Conference on Machine Learning, ICML 2016, New York City, NY, USA, June 19-24, 2016 (JMLR Workshop and Conference P...

    2016

  36. [44]

    Jianxiang Yu, Jiapeng Zhu, Hao Qian, Ziqi Liu, Zhiqiang Zhang, and Xiang Li

  37. [45]

    Xingtong Yu, Zechuan Gong, Chang Zhou, Yuan Fang, and Hui Zhang. 2025. SAMGPT: Text-free Graph Foundation Model for Multi-domain Pre-training and Cross-domain Adaptation. InProceedings of the ACM on Web Conference 2025, WWW 2025, Sydney, NSW, Australia, 28 April 2025- 2 May 20...

    2025

  38. [46]

    Xingtong Yu, Chang Zhou, Yuan Fang, and Xinming Zhang. 2024. Text-Free Multi-domain Graph Pre-training: Toward Graph Foundation Models.CoRR abs/2405.13934 (2024). arXiv:2405.13934 doi:10.48550/ARXIV.2405.13934

    2024 doi arXiv

  39. [47]

    Haonan Yuan, Qingyun Sun, Junhua Shi, Xingcheng Fu, Bryan Hooi, Jianxin Li, and Philip S. Yu. 2025. How Much Can Transfer? BRIDGE: Bounded Multi- Domain Graph Foundation Model with Generalization Guarantees. InForty- second International Conference on Machine Learning, ICML 20...

    2025

  40. [48]

    arXiv:2505.12027 doi:10.48550/ARXIV.2505.12027

    Relation-Aware Graph Foundation Model.CoRRabs/2505.12027 (2025). arXiv:2505.12027 doi:10.48550/ARXIV.2505.12027

    2025 doi arXiv

  41. [49]

    Jianan Zhao, Hesham Mostafa, Mikhail Galkin, Michael Bronstein, Zhaocheng Zhu, and Jian Tang. 2024. Graphany: A foundation model for node classification on any graph.arXiv preprint arXiv:2405.20445(2024)

    2024 arXiv

  42. [50]

    Jitao Zhao, Yi Wang, Yawen Li, Dongxiao He, Di Jin, Zhiyong Feng, and Weixiong Zhang. 2026. Towards Graph Foundation Model: Node Feature Transfer Invariant Modeling on General Graphs. InProceedings of the ACM Web Conference 2026, WWW 2026, Dubai, United Arab Emirates, original...

    2026 doi

  43. [51]

    Jie Zhou, Ganqu Cui, Shengding Hu, Zhengyan Zhang, Cheng Yang, Zhiyuan Liu, Lifeng Wang, Changcheng Li, and Maosong Sun. 2020. Graph neural networks: A review of methods and applications.AI open1 (2020), 57–81

    2020

  44. [52]

    Jitao Zhao, Di Jin, Meng Ge, Lianze Shan, Xin Wang, Dongxiao He, and Zhiyong Feng. 2024. FUG: Feature-Universal Graph Contrastive Pre-training for Graphs with Diverse Node Features. InThe Thirty-eighth Annual Conference on Neural Information Processing Systems

    2024

  45. [56]

    Chenyi Zi, Haihong Zhao, Xiangguo Sun, Yiqing Lin, Hong Cheng, and Jia Li. 2024. ProG: A Graph Prompt Learning Benchmark. InAdvances in Neural Information Processing Systems 37: Annual Conference on Neural Information Processing Systems 2024, NeurIPS 2024, Vancouver, BC, Canad...

    2024

  46. [2015]

    InProceedings of the 38th International ACM SIGIR Conference on Research and Development in Information Retrieval, Santiago, Chile, August 9-13, 2015

    Image-Based Recommendations on Styles and Substitutes. InProceedings of the 38th International ACM SIGIR Conference on Research and Development in Information Retrieval, Santiago, Chile, August 9-13, 2015. ACM, 43–52. https: //doi.org/10.1145/2766462.2767755

    2015

  47. [2025]

    RiemannGFM: Learning a Graph Foundation Model from Riemannian Geometry. InProceedings of the ACM on Web Conference 2025, WWW 2025, Sydney, NSW, Australia, 28 April 2025- 2 May 2025, Guodong Long, Michale Blumestein, Yi Chang, Liane Lewin-Eytan, Zi Helen Huang, and Elad Yom-Tov...

    2025

  48. [2026]

    LEDA: Latent Semantic Distribution Alignment for Multi-domain Graph Pre-training. InProceedings of the ACM Web Conference 2026, WWW 2026, Dubai, United Arab Emirates, originally scheduled for April 13-17, 2026, rescheduled for June 29 - July 3, 2026, Hakim Hacid, Yoelle Maarek...

    2026

Pith tools

Reviewed August 3, 2026 · model on record in the stance chip above.