Pith. sign in

REVIEW 5 major objections 5 minor 32 references

Graph Similarity Computation via Interpretable Neural Node Alignment

T0 review · 5 major / 5 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read The paper claims that a graph neural network with a differentiable Gumbel-Sinkhorn layer can predict graph edit distance and output a valid one-to-one node alignment, trained only on graph-level similarity scores.

desk verdict GNA gets hard one-to-one node alignment working without node-level labels and posts strong numbers, but the padding scheme quietly forbids deleting nodes from the smaller graph, and the novelty claim overstates GOTSim's prior one-to-one alignment. read the letter →

arxiv 2412.12185 v1 pith:O4F2M3C2 submitted 2024-12-13 cs.LG cs.AI

Jingjing Wang , Hongjie Zhu , Haoran Xie , Fu Lee Wang , Xiaoliang Xu , Yuxiang Wang This is my paper · ORCID
classification cs.LGcs.AI
keywords GraphSimilarityRepresentationModelInterpretabilityEditDistanceNodeAlignmentGumbel-SinkhornNeuralNetworkRetrieval
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

Graph edit distance (GED), the minimal cost of turning one graph into another, is NP-hard to compute exactly, and neural approximations usually hide the node-correspondence step that classical algorithms make visible. This paper proposes GNA, a model that predicts GED while also outputting a hard one-to-one alignment between nodes of the two graphs, without any node-level alignment labels. The authors relax the classical quadratic assignment problem to a linear one in a learned node-embedding space, then use a differentiable Gumbel-Sinkhorn module to produce a doubly stochastic permutation matrix. On three real-world graph datasets, they report that GNA reduces mean squared error by up to 16% and improves retrieval ranking metrics by up to 12% compared with existing neural baselines. The central message is that interpretability via bijective node alignment need not come at the cost of accuracy; in their experiments it helps prediction.

What carries the argument

The central mechanism is the relaxation of the quadratic assignment problem of GED to a linear assignment in node-embedding space, combined with Gumbel-Sinkhorn normalization to enforce a doubly stochastic, one-to-one permutation matrix in a differentiable way. The GIN encoder supplies node features that are assumed to encode subgraph and edge structure, so that the node-node cost matrix alone can stand in for the full adjacency-matching objective. The cost matrix and matching matrix are multiplied elementwise to produce the predicted edit distance, with the Sinkhorn iterations providing the hard, interpretable alignment that soft-alignment methods lack.

What would settle it

Compute exact GED and the exact optimal alignment for a sample of small graphs ($\le 10$ nodes) with A* search, then compare GNA's predicted permutation and predicted GED against those ground truths; if the model often returns a different one-to-one correspondence while still predicting an accurate similarity score, then the linear relaxation is capturing the right score for the wrong structural reason.

Watch

Extended reading notes

Core claim

GNA treats GED prediction as a prediction task with a differentiable bijective matching layer. Given two graphs, a three-layer GIN encoder produces node features that, the paper argues, capture both subgraph structure and edge information, which reduces the quadratic assignment underlying GED to a linear assignment over node embeddings. A costing module computes pairwise substitution costs, with padded nodes carrying graph-level embeddings to represent insertion and deletion. A matching module feeds transformed embeddings through Gumbel-Sinkhorn iterations to obtain a doubly stochastic, approximately binary permutation matrix that enforces one-to-one matching. The predicted GED is the elementwise product of the cost matrix and the matching matrix plus an NTN bias, trained with mean squared error against normalized GED. The paper claims this is the first method to predict graph similarity while producing one-to-one node matching without alignment ground truth, and reports superior accuracy on both similarity estimation and retrieval, with an ablation showing that both the bijection constraint and the add/delete costs matter.

Load-bearing premise

The whole approach rests on the assumption that the GIN node embeddings alone capture enough subgraph and edge structure that the linear node-assignment problem is a faithful substitute for the full quadratic assignment problem of GED; the paper asserts this but gives no proof or quantitative check.

Editorial extensions

If this is right

  • Neural GED predictors can be made interpretable without extra labels: the model's permutation matrix shows which nodes map to which, much like classical edit paths.
  • Enforcing bijective alignment improves accuracy on similarity regression and retrieval, not just transparency; the ablation attributes this to the Gumbel-Sinkhorn constraint.
  • Insertion and deletion costs can be read off the alignment: padded rows and columns identify nodes that must be added or deleted when graph sizes differ.
  • The same architecture applies to any domain-agnostic similarity task on undirected unweighted graphs where a GED or MCS score is the supervision signal.

Reading between the lines

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

  • Quantitative comparison of GNA's predicted alignments against exact optimal alignments on small graphs would directly test the linear relaxation; the paper's evaluation of alignment quality is mostly visual.
  • Because the model needs no alignment labels, it could transfer to settings such as knowledge graph or social network alignment where similarity scores exist but correspondence ground truth is scarce.
  • The Gumbel-Sinkhorn temperature, set to 0.1 in the experiments, controls the trade-off between differentiability and closeness to a true permutation; tuning it per task is a natural extension.
  • If the relaxation is faithful, the matching matrix could be reused as a differentiable proxy for the full edit path, potentially enabling end-to-end learning of edge-level edit operations as the paper's future-work direction suggests.
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

5 major / 5 minor

Summary. The paper proposes GNA, a graph neural network model that approximates Graph Edit Distance and also outputs a hard one-to-one node alignment, trained without node-alignment ground truth. A GIN encoder produces node embeddings; a costing module builds a node-pair edit-cost matrix; a Gumbel-Sinkhorn module generates a doubly stochastic matching matrix; and the element-wise product of matching and cost, plus a graph-level bias, is passed through a sigmoid and trained with MSE against normalized GED. Experiments on AIDS, Linux, and IMDB compare GNA with seven baselines on MAE, Accuracy, rank correlations, and P@10, and the paper includes ablation and visualization studies. Source code is released.

Significance. If the central claims hold, the paper would make a modest but useful contribution to interpretable graph similarity computation by showing that a differentiable Sinkhorn-based alignment layer can be trained end-to-end from GED targets alone and that the resulting alignment offers visual interpretability. The empirical comparison spans three standard benchmarks and seven baselines, the code is released, and the case studies in Section 5.6 are a genuine strength. The performance claim is plausible but is currently not fully supported by the reported tables, and the alignment module as formulated does not represent all edit operations permitted by classical GED, so the main interpretability claim needs technical revision.

major comments (5)
  1. [4.2, Eq. (6); 4.3, Eq. (9)] The padding scheme appends only |V2|-|V1| dummy rows to the smaller graph and never creates dummy columns. In the resulting |V2| x |V2| permutation matrix, every original G1 row must be assigned to a distinct G2 column, so no edit path that deletes a node of the smaller graph can be represented. Classical GED as defined in Section 3 permits deletions from either graph, and when deleting a small-graph node is cheaper than substituting it, the feasible set here excludes the optimal edit path. This contradicts the paper's repeated claim that the hard alignment is consistent with classical GED and weakens the interpretability interpretation in Section 5.6. The model needs either a rectangular matching formulation with dummy columns on both sides, or the claims should be restricted to a substitution-only alignment.
  2. [4.4, Eq. (15)] The normalization denominator max(|V1|,|V2|)+max(|E1|,|E2|) is not an upper bound on GED under standard unit-cost edits. For example, take two 10-node graphs whose edge sets are disjoint and have 20 edges each: transforming one into the other requires at least 40 edge deletions and insertions, while the denominator equals 10 + 20 = 30. The normalized ground truth can therefore exceed 1, while the prediction in Eq. (14) is confined to [0,1] by the sigmoid, so the training target can be unreachable. The normalization should be changed to a true upper bound, or a proof of the bound must be supplied.
  3. [Abstract; Table 2] The abstract's headline claim of 'up to 16% reduction in the Mean Squared Error' is not supported anywhere in the paper: Table 2 reports MAE, Accuracy, rho, tau, and P@10, but no MSE values or MSE improvements are given. Either add an MSE column and report the reduction, or revise the abstract so the quantitative claim matches the metrics actually reported.
  4. [5.3, Table 2] No error bars, standard deviations, or statistical significance tests are reported, yet some of the claimed improvements are small or in the opposite direction (e.g., Linux rho: GNA 0.964 vs GEDGNN 0.963; IMDB MAE: GNA 1.425 vs GEDGNN 1.398, where GEDGNN is better). Without repeated runs or paired tests, the claim that GNA outperforms the baselines across all metrics is not statistically established. Please add repeated-run statistics and, ideally, paired significance tests.
  5. [4.3, Eqs. (10)-(12); 5.6] The Gumbel-Sinkhorn iterations with temperature tau=0.1 produce a continuous doubly stochastic matrix, not a binary {0,1} permutation matrix. The paper does not describe any discretization, rounding, or straight-through estimator that would produce the hard one-to-one alignment claimed as a contribution and displayed in Figure 4. As written, the matching module outputs a soft alignment at inference, so both the 'hard alignment' claim and the interpretability guarantee are not established. The authors should specify how the binary matrix is extracted or adjust the claims to soft alignments.
minor comments (5)
  1. [4.3, Eq. (9)] Equation (9) writes LRL_theta(H1) while the text says LRL is applied to the padded H*1; to obtain a |V2| x |V2| matching matrix, the padded embedding must be used. Please correct the notation.
  2. [4.4, Eq. (15)] The sentence 'the denominator is the original GED value' is incorrect; the denominator is the normalization factor max(|V1|,|V2|)+max(|E1|,|E2|). Please correct the wording.
  3. [1 vs 2.2] The introduction lists GOTSim among soft-alignment methods, while Section 2.2 credits GOTSim with deriving one-to-one node alignment from a similarity matrix and an addition/deletion cost matrix. The novelty claim of being 'first' to provide one-to-one node matching should be reconciled with this description of GOTSim.
  4. [5.4] The ablation text says that removing Gumbel-Sinkhorn is intended to 'relax the bijection limitation' but then reports a performance decline; please clarify what 'relax' means here and report numeric ablation values rather than only figures.
  5. [5.3 and 2.1] There are typos including 'similarity similarity task' in Section 5.4 and 'Howvever' in Section 2.1; these should be corrected.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: GNA is trained end-to-end on exact GED ground truth and evaluated on held-out splits; the node alignment is a latent variable, not a fitted target.

full rationale

The paper's derivation chain is self-contained. The model parameters are optimized by minimizing the MSE in Eq. (16) between the predicted normalized score in Eq. (14) and the ground-truth normalized GED in Eq. (15) computed by exact GED. The matching matrix in Eq. (9) is produced by the Gumbel-Sinkhorn module from learned embeddings without node-alignment ground truth; it is not fitted to any label that the prediction later claims to discover. No equation defines a 'prediction' in terms of its own input, and no fitted parameter is renamed as a prediction. The claim that GIN node embeddings encapsulate subgraph and edge information is an assumption about representation quality, not a circular reduction. The reviewer's concern about padding only the smaller graph with dummy rows (Eq. 6) is a modeling-correctness issue: it may exclude deletion edit paths and bias the GED estimate, but that does not make the output equivalent to the training target or to the cost matrix by construction. There are no load-bearing self-citations: references to prior work are used for standard architectural choices (e.g., GIN, SimGNN-style branches) and dataset construction, not to justify the central prediction claim. The central result is empirically validated against exact GED on held-out test pairs, so the work is not circular.

Assumptions & free parameters 5 free parameters · 4 assumptions · 0 invented entities

The central claim rests on several unverified modeling assumptions: the linear relaxation equivalence, the padding strategy, and the Sinkhorn-Gumbel approximation. These are pragmatic choices asserted rather than derived. The free parameters are architecture and training hyperparameters, not new scientific constants. No new physical or conceptual entities are introduced, so the invented entities list is empty.

free parameters (5)
  • Gumbel-Sinkhorn temperature tau = 0.1
    Controls the smoothness versus discreteness of the doubly stochastic alignment matrix; chosen by hand and not tuned per dataset in a reported sweep.
  • Number of bilinear cost layers l = 16
    Depth of the stacked bilinear interaction layers in the costing module; selected by validation performance.
  • GIN hidden dimensions = 32, 64, 128
    Channel sizes for the three GIN layers, taken from GEDGNN and used without a reported sensitivity analysis.
  • Learning rate = 0.001
    Adam optimizer learning rate, a standard choice for this model family.
  • Weight decay = 5e-4
    Regularization coefficient used during training; standard value.
assumptions (4)
  • domain assumption GIN node embeddings fully capture subgraph structure and edge information, making the linear node-assignment relaxation (Eq. 3) equivalent to the quadratic assignment problem (Eq. 1/2).
    Stated in Section 4.1: 'the node features generated by the GIN encoder encapsulate both the subgraph structure and edge feature information.' No formal bound or empirical verification of this equivalence is provided.
  • ad hoc to paper Padding the smaller graph with the graph-level mean-pooling embedding yields a valid add/delete cost model.
    Section 4.2 defines the initial features of padded nodes as the mean-pooled graph embedding, assuming this captures insertion and deletion costs; alternative padding strategies are not evaluated.
  • domain assumption Iterative row and column normalization with Gumbel noise converges to a doubly stochastic matrix that approximates the optimal permutation matrix.
    Equations 10-12 in Section 4.3 invoke the Sinkhorn-Gumbel construction, assuming that a finite number of iterations at temperature tau=0.1 is a good approximation to the infinite limit.
  • ad hoc to paper The sigmoid of match*cost plus bias (Eq. 14) can be trained by MSE against the normalized GED target (Eq. 15), and the target lies in [0,1].
    Section 4.4 claims the denominator is 'the original GED value' and ensures a [0,1] score, but the denominator is max(|V|)+max(|E|), which does not necessarily bound the graph edit distance; the mismatch between sigmoid output and linear normalization is not analyzed.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Graph Similarity Computation via Interpretable Neural Node Alignment." pith.science (2026). https://pith.science/paper/O4F2M3C2

@misc{pith2026241212185,
  author       = {Pith},
  title        = {Pith review of: Graph Similarity Computation via Interpretable Neural Node Alignment},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/O4F2M3C2}},
  note         = {Machine review of arXiv:2412.12185}
}
read the original abstract

\Graph similarity computation is an essential task in many real-world graph-related applications such as retrieving the similar drugs given a query chemical compound or finding the user's potential friends from the social network database. Graph Edit Distance (GED) and Maximum Common Subgraphs (MCS) are the two commonly used domain-agnostic metrics to evaluate graph similarity in practice. Unfortunately, computing the exact GED is known to be a NP-hard problem. To solve this limitation, neural network based models have been proposed to approximate the calculations of GED/MCS. However, deep learning models are well-known ``black boxes'', thus the typically characteristic one-to-one node/subgraph alignment process in the classical computations of GED and MCS cannot be seen. Existing methods have paid attention to approximating the node/subgraph alignment (soft alignment), but the one-to-one node alignment (hard alignment) has not yet been solved. To fill this gap, in this paper we propose a novel interpretable neural node alignment model without relying on node alignment ground truth information. Firstly, the quadratic assignment problem in classical GED computation is relaxed to a linear alignment via embedding the features in the node embedding space. Secondly, a differentiable Gumbel-Sinkhorn module is proposed to unsupervised generate the optimal one-to-one node alignment matrix. Experimental results in real-world graph datasets demonstrate that our method outperforms the state-of-the-art methods in graph similarity computation and graph retrieval tasks, achieving up to 16\% reduction in the Mean Squared Error and up to 12\% improvement in the retrieval evaluation metrics, respectively.

Figures

Figures reproduced from arXiv: 2412.12185 by the authors.

Figure 1
Figure 1. The framework of the GNA. The proposed interpretable neural node alignment model, GNA, which comprises three components: 1) Embedding Module and 2) Costing and Matching Module, 3) Prediction, as illustrated in [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Ablation study result. The top subfigures denotes on the original test set, the bottom ones are conducted on the filtered dataset. In this section, we evaluate the affect of the two key components in GNA: bijection in the matching module and delete/add cost in the cost module. w/o Add/Delete Cost refers to removing the add and deleting cost; w/o Gumbel￾sinkhorn refers to removing the gumbel-sinkhorn algorithm to rel… view at source ↗
Figure 3
Figure 3. Visualizations of graph ranking result on AIDS. Nodes with the same color have the same label. approximates to the ground truths. Furthermore, we obtain the same conclusions as shown in [PITH_FULL_IMAGE:figures/full_fig_p010_3.png] view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: A graph matching case study for GNA on AIDS datasets. Heatmap denotes the final matching matrix at the node-level based on the node features generated by matching module. The depth of the color in the heat map indicates the weight of the node matching. The deeper the c…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

32 extracted references · 26 canonical work pages

  1. [1]

    Bunke, On a relation between graph edit distance and maximum common subgraph, Pattern recognition letters 18 (8) (1997) 689–694

    H. Bunke, On a relation between graph edit distance and maximum common subgraph, Pattern recognition letters 18 (8) (1997) 689–694

    work page 1997

  2. [2]

    Neuhaus, K

    M. Neuhaus, K. Riesen, H. Bunke, Fast suboptimal algorithms for the computation of graph edit distance, in: Structural, Syntactic, and Statistical Pattern Recognition: Joint IAPR International Workshops, SSPR 2006 and SPR 2006, Hong Kong, China, August 17-19, 2006. Proceedings, Springer, 2006, pp. 163–172

    work page 2006

  3. [3]

    K.Riesen,H.Bunke,Approximategrapheditdistancecomputationbymeansofbipartitegraphmatching,ImageandVisioncomputing27(7) (2009) 950–959

    work page 2009

  4. [4]

    Fankhauser, K

    S. Fankhauser, K. Riesen, H. Bunke, Speeding up graph edit distance computation through fast bipartite matching, in: Graph-Based Representations in Pattern Recognition: 8th IAPR-TC-15 International Workshop, GbRPR 2011, Münster, Germany, May 18-20, 2011. Proceedings 8, Springer, 2011, pp. 102–111

    work page 2011

  5. [5]

    C. Piao, T. Xu, X. Sun, Y. Rong, K. Zhao, H. Cheng, Computing graph edit distance via neural graph matching, Proceedings of the VLDB Endowment 16 (8) (2023) 1817–1829

    work page 2023

  6. [6]

    35, 2021, pp

    H.-R.Tan,C.Wang,S.-T.Wu,T.-Q.Wang,X.-Y.Zhang,C.-L.Liu,Proxygraphmatchingwithproximalmatchingnetworks,in:Proceedings of the AAAI conference on artificial intelligence, Vol. 35, 2021, pp. 9808–9815

    work page 2021

  7. [7]

    K. Zhao, S. Tu, L. Xu, Ia-gm: A deep bidirectional learning method for graph matching, in: Proceedings of the AAAI conference on artificial intelligence, Vol. 35, 2021, pp. 3474–3482

    work page 2021

  8. [8]

    M. Fey, J. E. Lenssen, C. Morris, J. Masci, N. M. Kriege, Deep graph matching consensus, arXiv preprint arXiv:2001.09621 (2020)

    arXiv 2020

Show all 32 references
  1. [9]

    I. Roy, V. S. B. R. Velugoti, S. Chakrabarti, A. De, Interpretable neural subgraph matching for graph retrieval, in: Proceedings of the AAAI conference on artificial intelligence, Vol. 36, 2022, pp. 8115–8123

    2022

  2. [10]

    K. D. Doan, S. Manchanda, S. Mahapatra, C. K. Reddy, Interpretable graph similarity computation via differentiable optimal alignment of nodeembeddings,in:Proceedingsofthe44thInternationalACMSIGIRConferenceonResearchandDevelopmentin InformationRetrieval, 2021, pp. 665–674

    2021

  3. [11]

    L. Chen, Z. Gan, Y. Cheng, L. Li, L. Carin, J. Liu, Graph optimal transport for cross-domain alignment, in: International Conference on Machine Learning, PMLR, 2020, pp. 1542–1553

    2020

  4. [12]

    Petric Maretic, M

    H. Petric Maretic, M. El Gheche, G. Chierchia, P. Frossard, Got: an optimal transport framework for graph comparison, Advances in Neural Information Processing Systems 32 (2019)

    2019

  5. [13]

    R.Jonker,T.Volgenant,Ashortestaugmentingpathalgorithmfordenseandsparselinearassignmentproblems,in:DGOR/NSOR:Papersof the16thAnnualMeetingofDGORinCooperationwithNSOR/Vorträgeder16.JahrestagungderDGORzusammenmitderNSOR,Springer, 1988, pp. 622–622

    1988

  6. [14]

    Z. Lou, J. You, C. Wen, A. Canedo, J. Leskovec, et al., Neural subgraph matching, arXiv preprint arXiv:2007.03092 (2020)

    2020 arXiv

  7. [15]

    A.Grover,J.Leskovec,node2vec:Scalablefeaturelearningfornetworks,in:Proceedingsofthe22ndACMSIGKDDinternationalconference on Knowledge discovery and data mining, 2016, pp. 855–864

    2016

  8. [16]

    Perozzi, R

    B. Perozzi, R. Al-Rfou, S. Skiena, Deepwalk: Online learning of social representations, in: Proceedings of the 20th ACM SIGKDD international conference on Knowledge discovery and data mining, 2014, pp. 701–710

    2014

  9. [17]

    Y.Bai,H.Ding,S.Bian,T.Chen,Y.Sun,W.Wang,Simgnn:Aneuralnetworkapproachtofastgraphsimilaritycomputation,in:Proceedings of the twelfth ACM international conference on web search and data mining, 2019, pp. 384–392

    2019

  10. [18]

    Y. Bai, H. Ding, K. Gu, Y. Sun, W. Wang, Learning-based efficient graph similarity computation via multi-scale convolutional set matching, in: Proceedings of the AAAI conference on artificial intelligence, Vol. 34, 2020, pp. 3219–3226

    2020

  11. [19]

    3835–3845

    Y.Li,C.Gu,T.Dullien,O.Vinyals,P.Kohli,Graphmatchingnetworksforlearningthesimilarityofgraphstructuredobjects,in:International conference on machine learning, PMLR, 2019, pp. 3835–3845. Wang J et al.:Preprint submitted to Elsevier Page 11 of 12

    2019

  12. [20]

    Zheng, H

    J. Zheng, H. Chen, Q. Ma, Cross-domain named entity recognition via graph matching, arXiv preprint arXiv:2408.00981 (2024)

    2024 arXiv

  13. [21]

    H. Peng, J. Li, Q. Gong, Y. Ning, S. Wang, L. He, Motif-matching based subgraph-level attentional convolutional network for graph classification, in: Proceedings of the AAAI conference on artificial intelligence, Vol. 34, 2020, pp. 5387–5394

    2020

  14. [22]

    Z. Zhao, P. Qian, X. Yang, Z. Zeng, C. Guan, W. L. Tam, X. Li, Semignn-ppi: Self-ensembling multi-graph neural network for efficient and generalizable protein-protein interaction prediction, arXiv preprint arXiv:2305.08316 (2023)

    2023 arXiv

  15. [23]

    J. Tang, W. Zhang, J. Li, K. Zhao, F. Tsung, J. Li, Robust attributed graph alignment via joint structure learning and optimal transport, in: 2023 IEEE 39th International Conference on Data Engineering (ICDE), IEEE, 2023, pp. 1638–1651

    2023

  16. [24]

    Y.Ye,X.Lian,M.Chen,Efficientexactsubgraphmatchingviagnn-basedpathdominanceembedding,ProceedingsoftheVLDBEndowment 17 (7) (2024) 1628–1641

    2024

  17. [25]

    D. Yao, Y. Gu, G. Cong, H. Jin, X. Lv, Entity resolution with hierarchical graph attention networks, in: Proceedings of the 2022 International Conference on Management of Data, 2022, pp. 429–442

    2022

  18. [26]

    L.Yang,L.Zou,Noah:Neural-optimizeda*searchalgorithmforgrapheditdistancecomputation,in:2021IEEE37thInternationalConference on Data Engineering (ICDE), IEEE, 2021, pp. 576–587

    2021

  19. [27]

    Raveaux, J.-C

    R. Raveaux, J.-C. Burie, J.-M. Ogier, A graph matching method and a graph matching distance based on subgraph assignments, Pattern Recognition Letters 31 (5) (2010) 394–406

    2010

  20. [28]

    K. Xu, W. Hu, J. Leskovec, S. Jegelka, How powerful are graph neural networks?, arXiv preprint arXiv:1810.00826 (2018)

    2018 arXiv

  21. [29]

    Hamilton, Z

    W. Hamilton, Z. Ying, J. Leskovec, Inductive representation learning on large graphs, Advances in neural information processing systems 30 (2017)

    2017

  22. [30]

    W. Zhuo, G. Tan, Efficient graph similarity computation with alignment regularization, Advances in Neural Information Processing Systems 35 (2022) 30181–30193

    2022

  23. [31]

    H.Xu,Z.Duan,Y.Wang,J.Feng,R.Chen,Q.Zhang,Z.Xu,Graphpartitioningandgraphneuralnetworkbasedhierarchicalgraphmatching for graph similarity computation, Neurocomputing 439 (2021) 348–362

    2021

  24. [32]

    J. Bai, P. Zhao, Tagsim: Type-aware graph similarity learning and computation, Proceedings of the VLDB Endowment 15 (2) (2021). Wang J et al.:Preprint submitted to Elsevier Page 12 of 12

    2021

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.