Pith. sign in

REVIEW 3 major objections 6 minor 1 cited by

A Self-Explainable Heterogeneous GNN for Relational Deep Learning

T0 review · 3 major / 6 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read MPS-GNN learns meta-paths whose informative content is aggregate statistics over occurrences, making predictions depend only on the meta-path-induced subgraph.

desk verdict A genuinely useful extension of MP-GNN to count-based meta-path statistics, with real experimental gains, but the scoring function's per-node weights carry an unverified materializability assumption that should be pinned down before the main claim is taken as settled. read the letter →

arxiv 2412.00521 v2 pith:GBQRRMW3 submitted 2024-11-30 cs.LG cs.DB

Francesco Ferrini , Antonio Longa , Andrea Passerini , Manfred Jaeger This is my paper · ORCID
classification cs.LGcs.DB MSC 68T0768R10
keywords meta-pathlearningheterogeneousgraphneuralnetworksrelationaldeepself-explainableGNNweightedmulti-instanceclassificationaggregatestatisticscounts-of-countsdatabases
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

The paper introduces MPS-GNN (Meta-Path Statistics GNN), a heterogeneous graph neural network that discovers meta-paths—sequences of relations—whose predictive value lies in aggregate statistics over their occurrences, such as counts-of-counts patterns ('at least two exempt prescriptions, each containing at least two medications'), rather than in the mere existence of a single occurrence. The authors argue that this is the right inductive bias for relational databases, where labels depend on statistics computed over related rows. They claim the method finds these meta-paths automatically, scales linearly in the number of relations, and is genuinely self-explainable because the trained network can only access the subgraph induced by the identified meta-paths. Experiments on synthetic count-based scenarios and on medical, geographic, and Formula 1 databases report large F1 gains over non-relational, heterogeneous, and meta-path baselines.

What carries the argument

The load-bearing object is the node feature function $f(v,r,\Theta,w)=\Theta^\top x_v$ when $v$ has no $r$-neighbors, and $f(v,r,\Theta,w)=\Theta^\top x_v \sum_{u\in N^r_v} w_u$ otherwise, with per-node weights $w_u\in[0,1]$; the sum over the $r$-neighborhood is what lets the score count occurrences instead of checking existence. Relation selection minimizes $L(r)=\min_{\Theta,w}\sum_{B^+\in S^+, B^-\in S^-}\sigma(F(B^-)-F(B^+))$, and newly created bags with weights from Equations (5)–(6) propagate informativeness to the next iteration. The MPS-GNN layer $h^{(l+1)}_v=\sigma(W^{(l)}_0 h^{(l)}_v + W^{(l)}_{\mathrm{neigh}}\sum_{u\in N^{r_{L-l}}_v}h^{(l)}_u + W^{(l)}_1 h^{(0)}_v)$ includes a skip connection from the input features at every layer, which the ablation shows is needed to keep node attributes available when statistics are computed.

What would settle it

Build a synthetic database whose label depends on at least $c$ occurrences of a known meta-path, and add a decoy relation that connects each target node to a unique dummy node carrying a one-hot identifier. If the scoring function gives the decoy relation a lower loss than the correct one, or if the trained MPS-GNN's predictions change when the decoy subgraph is modified, then the central claim that the weights represent materializable features is falsified.

Watch

Extended reading notes

Core claim

The central claim is that class membership in relational data can be determined by learnable statistics over meta-path realizations, and that such meta-paths can be identified by a greedy, local search without user supervision. The search keeps a growing meta-path prefix; at each step it scores every candidate next relation by solving a weighted multi-instance classification problem whose node weights act as 'putative features' that later steps materialize as real features. A sum aggregation over neighbors replaces the existential max of the predecessor method, so multiple occurrences contribute to the score. The final MPS-GNN uses only the subgraph spanned by the occurrences of the selected meta-paths, which makes the meta-paths sufficient explanations by construction; necessity is then verified by removing occurrences and observing a drop in F1 and a rise in predicted-probability distance.

Load-bearing premise

The search assumes that the per-node weights learned to separate positive and negative bags stand for features that can be materialized from node attributes or from meta-path extensions; if those weights simply memorize node identities, the relation scores can point the greedy search at uninformative meta-paths.

Editorial extensions

If this is right

  • Relational databases with many tables can be handled without a domain expert listing the relevant meta-paths, because the scoring function selects relations by their potential to support aggregate features.
  • The search cost for a meta-path of length $L$ drops from testing all $|R|^L$ paths to $O(|R|\cdot L)$ scoring steps, at fixed beam size.
  • The discovered meta-paths double as model-level explanations: predictions cannot change when parts of the graph outside the meta-path-induced subgraph change, and removing meta-path occurrences measurably degrades performance.
  • Count-based synthetic tasks—where at least $c$ occurrences of a length-$l$ meta-path decide the label—are solved near-optimally, while the existential-only predecessor fails on the same tasks.
  • On EICU, MONDIAL, and ErgastF1 the method reports the highest F1 among the compared models, with the identified meta-paths matching domain intuition about vital signs, language/border/ethnic-group information, and standings.

Reading between the lines

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

  • Because the scoring function is already a weighted multi-instance regression in disguise, the same machinery should extend to multiclass and regression targets by replacing scalar weights with vector-valued versions; the authors note this extension but leave it to future work.
  • The discovered meta-paths could be compiled directly into SQL aggregate features (COUNT, SUM, AVG over joined tables), offering a testable bridge between the learned explanations and classical relational feature engineering.
  • A caveat the paper itself states: the scoring function relies on a well-connected graph, and disconnected neighborhoods require supernode preprocessing; datasets with sparse connectivity may need that step before the guarantees apply.
  • One can test the 'putative feature' assumption directly by checking whether the learned per-node weights correlate with node attributes on held-out bags; if they instead memorize node identities, the greedy search could be misled in larger graphs.
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 / 6 minor

Summary. The paper proposes MPS-GNN, a heterogeneous GNN for relational databases that automatically learns meta-paths whose predictive power derives from aggregate statistics over multiple occurrences (e.g., counts and counts-of-counts), rather than merely from the existence of a single occurrence. The method extends MP-GNN by replacing max aggregation with sum aggregation in a weighted multi-instance scoring function, and by iteratively constructing meta-paths through a greedy, local search guided by a surrogate loss. The final classifier is a multi-relational GNN restricted to the learned meta-path-induced subgraph, which the paper argues makes the model self-explainable by construction. Experimental evaluation on eight synthetic count-based scenarios and three real-world databases (EICU, MONDIAL, ErgastF1) reports state-of-the-art F1 scores, and the identified meta-paths are shown to be faithful explanations via deletion-based necessity tests.

Significance. If the claims hold, the paper addresses a real limitation of existing meta-path learning methods in relational deep learning: moving beyond existential quantification to aggregate statistics. The method is computationally efficient relative to full meta-path enumeration, the code is released, and the synthetic experiments provide a clear falsifiable testbed for the central claim. The self-explainability framing is also valuable, as the final model's predictions are, by construction, functions only of the selected meta-path subgraphs. These strengths make the paper a potentially useful contribution to the community.

major comments (3)
  1. [Section 4.2, Eq. (2)-(4)] The per-node weights w_u in Eq. (2) are unconstrained free parameters, and the loss L(r) in Eq. (4) is minimized over all such weights. This allows w_u to memorize node identities: if each positive bag contains an r-successor not present in any negative bag, setting those w_u to 1 and all others to 0 drives the loss to zero, even when the r-neighborhood is uninformative for any realizable feature. The paper does not constrain w_u to be a function of node attributes or of meta-path extensions, nor does it analyze when the greedy search could get stuck on a spurious first relation. The toy example in Section 4.2 is an instance of this: relation b is selected solely because w_u separates the bags, while the b-only meta-path is non-discriminative; the method succeeds only because a later relation d materializes the required count statistic. Please either (a) constrain or regularize w_u so that they correspond to materializable features, or (b) provide an empirical or theoretical analysis showing that the validation-based F1 check in Algorithm 1 reliably prevents the selection of meta-paths whose apparent informativeness cannot be materialized. As written, the central claim that MPS-GNN automatically identifies meta-paths for aggregate statistics is conditional on an unverified assumption about the scoring function.
  2. [Section 5.3 and Eq. (for necessity)] The necessity metric Nec = 1/N Σ (p_v(G) − p_v(G′)) is computed by randomly deleting meta-path occurrences and measuring the change in predicted probabilities. Because the model's computational graph only contains the meta-path subgraph, any deletion changes predictions, so this test confirms sensitivity to the explanation but does not specifically verify that predictions depend on aggregate counts rather than on the existence of any occurrence. A targeted experiment that reduces occurrence counts from c to c−1 while preserving at least one occurrence (or that compares predictions under count perturbations versus existence-preserving perturbations) would directly support the aggregate-statistics claim. Without such a test, the self-explainability evidence is weaker than the synthetic experiments, which already address this distinction.
  3. [Section 4.2.1] The complexity claim that MPS-GNN 'scales linearly in the number of relations and nodes' is imprecise. Minimizing Eq. (4) for each candidate relation involves optimizing over a vector w with one entry per node reachable via that relation, which is O(|V|) parameters per relation per iteration, giving a worst-case cost of O(L·|R|·|V|) for the full meta-path construction. This is still far cheaper than enumerating all meta-paths, but the 'linear' statement should be qualified, and the per-iteration optimization cost should be stated explicitly so that readers can assess scalability on large relational databases.
minor comments (6)
  1. [Section 1] The sentence 'Existing approaches for heterogeneous GNNs either rely on domain experts to provide relevant meta-paths a priori, or attempt to learn them from data by assigning different weights to various relations, a solution that fails to scale with the number of candidate relations' is grammatically incomplete; consider rephrasing to clarify which solution fails to scale.
  2. [Section 4.2 and Algorithm 1] The stopping criterion in Algorithm 1, 'if min_r L(r) ≥ η Linit(r)', uses a quantity Linit(r) that is not defined in the algorithm or in the surrounding text; please define it and specify how the 30% improvement threshold is chosen and how sensitive the results are to it.
  3. [Section 5.2 / Appendix A.3] The preprocessing step that clusters rows of auxiliary tables to improve connectivity is described only briefly; its potential effect on the results (e.g., information leakage or sensitivity to the clustering method) is not discussed. Please add a sentence or an ablation assessing this.
  4. [Appendix A.8, Table 13] The RDL row for rel-f1-top3 reports a standard deviation of 0.7, which appears to be a typo (likely 0.07); please correct it.
  5. [Section 2] The claim that MPS-GNN is 'the first truly self-explainable GNN designed for relational deep learning applications' is too strong without a more thorough survey; please soften it to 'to the best of our knowledge' and add references to any prior self-explainable heterogeneous GNNs or meta-path-based explanation methods.
  6. [General] There are numerous typos and spacing errors throughout the text (e.g., 'behindMP-GNN', 'in both synthetic and real-world scenario', missing spaces before citations). A careful proofreading pass is recommended.

Circularity Check

1 steps flagged · score 2.0 of 10

No significant circularity: the meta-path selection and empirical F1 evaluation are self-contained; the only definitional reduction is the sufficiency half of the self-explainability claim, which holds by construction because the explanation is defined as the subgraph the model actually uses.

  1. self definitional [Section 4.5, 'MPS-GNN is a self-explainable model'; cf. Eq. (7) and Section 5.3]
    "By relying on meta-paths for its predictions, MPS-GNN is a self-explainable GNN model. The scoring function serves as the detector, identifying relevant meta-paths, while the network built using them acts as the classifier. By construction, the network can only access the meta-path induced subgraph, making it strictly sufficient by construction (no changes outside the meta-path induced graph affect the prediction)."

    The paper defines the explanation as the meta-path-induced subgraph and constructs the GNN (Eq. 7) so that each layer aggregates only over neighbors reachable via the relations in the selected meta-path. Therefore the 'strict sufficiency' property—that changes outside the explanation do not affect the prediction—is entailed by the chosen definition of explanation rather than demonstrated by an independent test. Section 5.3 then presents this constructional tautology as part of the evidence for Q3 ('MPS-GNN is a self-explainable method'), with only the necessity component measured empirically. The sufficiency half of the self-explainability claim thus reduces to the architecture by definition, not to a separately derived result.

full rationale

The paper's central derivation chain is self-contained: relation scoring in Section 4.2 optimizes the loss in Eq. (4), meta-paths are extended greedily, and the final MPS-GNN is trained and evaluated with held-out F1 scores against external baselines. The fitted per-node weights w_u in Eq. (2) are used only as a scoring signal for potential informativeness, and the reported F1 results come from training the actual GNN on the selected meta-path, so the empirical claims are not forced by the scoring fit. The main identified reduction is the sufficiency component of self-explainability: because the explanation is defined as the subgraph the model uses, sufficiency is true by construction. This does not undermine the independent necessity experiments or the comparative empirical results, but it does make the 'inherently sufficient' claim definitional rather than an empirical finding. I therefore find no harmful or load-bearing circularity beyond this minor definitional tautology, and assign a score of 2.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

The central claim rests on the greedy and local search heuristic, the assumption that per-node scoring weights can be materialized as features, and the standard relational-to-graph modeling assumption. No new physical entities are introduced. Key hyperparameters (max meta-path length, stopping threshold, beam size) are set by hand.

free parameters (3)
  • Maximum meta-path length LMAX = 4
    Set by hand for all experiments; bounds the depth of the greedy search and the number of MPS-GNN layers.
  • Stopping threshold eta = 0.7
    Controls when meta-path construction terminates in Algorithm 1; authors claim robustness but do not show a sensitivity analysis.
  • Beam size K = 3
    Number of meta-paths explored in the beam search; chosen for comparability with MP-GNN.
assumptions (4)
  • ad hoc to paper The greedy, local construction of meta-paths, extending one relation at a time based on a surrogate weighted multi-instance loss, will identify globally informative meta-paths.
    No theoretical guarantee is provided; the method is justified by analogy to decision trees and by experiments.
  • ad hoc to paper Per-node weights w_u optimized in the scoring function (Eq. 2) can be materialized as features computable from node attributes or from further meta-path extensions.
    The paper assumes this to connect scoring to actual features, but w_u may simply memorize node identities.
  • domain assumption A relational database can be faithfully represented as a heterogeneous graph, possibly after clustering rows into supernodes, without losing task-relevant information.
    Standard relational deep learning assumption; the clustering preprocessing in Appendix A.3 is a strong additional assumption.
  • domain assumption Binary node classification labels can be separated by linear discriminants over aggregate meta-path statistics.
    The weighted multi-instance formulation assumes this functional form for the scoring function.

how reviews work

0 comments
Cite this review

Pith. "Pith review of A Self-Explainable Heterogeneous GNN for Relational Deep Learning." pith.science (2026). https://pith.science/paper/GBQRRMW3

@misc{pith2026241200521,
  author       = {Pith},
  title        = {Pith review of: A Self-Explainable Heterogeneous GNN for Relational Deep Learning},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/GBQRRMW3}},
  note         = {Machine review of arXiv:2412.00521}
}
read the original abstract

Recently, significant attention has been given to the idea of viewing relational databases as heterogeneous graphs, enabling the application of graph neural network (GNN) technology for predictive tasks. However, existing GNN methods struggle with the complexity of the heterogeneous graphs induced by databases with numerous tables and relations. Traditional approaches either consider all possible relational meta-paths, thus failing to scale with the number of relations, or rely on domain experts to identify relevant meta-paths. A recent solution does manage to learn informative meta-paths without expert supervision, but assumes that a node's class depends solely on the existence of a meta-path occurrence. In this work, we present a self-explainable heterogeneous GNN for relational data, that supports models in which class membership depends on aggregate information obtained from multiple occurrences of a meta-path. Experimental results show that in the context of relational databases, our approach effectively identifies informative meta-paths that faithfully capture the model's reasoning mechanisms. It significantly outperforms existing methods in both synthetic and real-world scenario.

Figures

Figures reproduced from arXiv: 2412.00521 by the authors.

Figure 1
Figure 1. Left: Relational database schema for a medical domain. Right: Heterogeneous graph represen￾tation of (part of) the database. The highlighted subgraph shows a prototypical counts-of-counts pattern characterising positive patients, namely having at least two exempt prescriptions (represented by node fea￾ture T), each containing at least two medications. Existing heterogeneous GNNs struggle with these patterns as they … view at source ↗
Figure 2
Figure 2. Outline of greedy local meta-path construction [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Scoring the first two relations Similarly, relation b is scored. The loss function 4 now becomes σ(ΘT x1(w4 + w5 + w6) − Θ T x0(w2 + w3 + w4)), where x0 = x1 = (1, 0)T is the attribute vector representing the T value in a one-hot encoding. This loss can be brought arbitrarily close to zero, e.g. by w6 → 0, w5 → 0, w3 → 1, w2 → 1, and Θ = (Z, 0)T with Z ≫ 0. Consequently, relation b scores higher than relation a and … view at source ↗
Figures from the paper (7 more)
Figure 4
Figure 4. Figure 4: Bag generation and scoring of relation c. For solving the new classification task we score the candidate relations c and d. The right part of [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]
Figure 5
Figure 5. Figure 5: Scoring relation d. is not directly visible at this step in the meta-path construction process, but will be found by training an MPS-GNN on this meta-path, as detailed in sections 4.3 and 4.4. Check A.10 for a similar example with more complex node features 4.2.1 Time …
Figure 6
Figure 6. Figure 6: (Left) sample scenario. Nodes are labeled as positive if and only if they are the starting point of at least c = 3 instances of the l = 2 meta-path "grey node r−→ orange node s−→ green node". (Right) statistics of synthetic datasets, with |R| total number of relations,…
Figure 7
Figure 7. Figure 7: Extracted meta-paths for the three real world datasets. [PITH_FULL_IMAGE:figures/full_fig_p012_7.png]
Figure 8
Figure 8. Figure 8: Changes to F1 and posterior probability difference (necessity) when removing 25%, 50%, and 75% of the learned meta-path occurrences for the real-world datasets with MPS-GNN, dashed line, and MP-GNN, solid line [PITH_FULL_IMAGE:figures/full_fig_p013_8.png]
Figure 9
Figure 9. Figure 9: Changes to F1 and posterior probability difference (necessity) when removing 25%, 50%, and 75% of the learned meta-path occurrences for the real-world temporal tasks rel-f1-dnf and rel-f1-top3. A.9 Non-GNN models on real-world databases For the MONDIAL and ErgastF1 dat…
Figure 10
Figure 10. Figure 10: Toy example similar to Figure 1 with more complex features on prescription nodes. As depicted [PITH_FULL_IMAGE:figures/full_fig_p025_10.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Beyond Topological Self-Explainable GNNs: A Formal Explainability Perspective

    cs.LG 2025-02 accept novelty 7.0 of 10

    Self-explainable GNNs provably optimize minimal explanations that match prime implicants only for motif-based tasks, and a dual-channel extension recovers better rules.

Reference graph

Works this paper leans on

54 extracted references · 33 canonical work pages · cited by 1 Pith paper

  1. [1]

    Evaluating explainability for graph neural networks

    Chirag Agarwal, Owen Queen, Himabindu Lakkaraju, and Marinka Zitnik. Evaluating explainability for graph neural networks. Scientific Data, 10 0 (1): 0 144, 2023

    work page 2023

  2. [2]

    Graphframex: Towards systematic evaluation of explainability methods for graph neural networks

    Kenza Amara, Zhitao Ying, Zitao Zhang, Zhichao Han, Yang Zhao, Yinan Shan, Ulrik Brandes, Sebastian Schemm, and Ce Zhang. Graphframex: Towards systematic evaluation of explainability methods for graph neural networks. In Learning on Graphs Conference, pp.\ 44--1. PMLR, 2022

    work page 2022

  3. [3]

    GI nx-eval: Towards in-distribution evaluation of graph neural network explanations

    Kenza Amara, Mennatallah El-Assady, and Rex Ying. GI nx-eval: Towards in-distribution evaluation of graph neural network explanations. In XAI in Action: Past, Present, and Future Applications, 2023. URL https://openreview.net/forum?id=w6Qnoy2RXG

    work page 2023

  4. [4]

    Global explainability of gnns via logic combination of learned concepts

    Steve Azzolin, Antonio Longa, Pietro Barbiero, Pietro Lio, and Andrea Passerini. Global explainability of gnns via logic combination of learned concepts. In The Eleventh International Conference on Learning Representations, 2022

    2022

  5. [5]

    Reconsidering faithfulness in regular, self-explainable and domain invariant GNN s

    Steve Azzolin, Antonio Longa, Stefano Teso, and Andrea Passerini. Reconsidering faithfulness in regular, self-explainable and domain invariant GNN s. In The Thirteenth International Conference on Learning Representations, 2025. URL https://openreview.net/forum?id=kiOxNsrpQy

    work page 2025

  6. [6]

    Interaction networks for learning about objects, relations and physics

    Peter Battaglia, Razvan Pascanu, Matthew Lai, Danilo Jimenez Rezende, et al. Interaction networks for learning about objects, relations and physics. Advances in neural information processing systems, 29, 2016

    2016

  7. [7]

    Simple decision forests for multi-relational classification

    Bahareh Bina, Oliver Schulte, Branden Crawford, Zhensong Qian, and Yi Xiong. Simple decision forests for multi-relational classification. Decision Support Systems, 54 0 (3): 0 1269--1279, 2013

    work page 2013

  8. [8]

    Megnn: Meta-path extracted graph neural network for heterogeneous graph representation learning

    Yaomin Chang, Chuan Chen, Weibo Hu, Zibin Zheng, Xiaocong Zhou, and Shouzhi Chen. Megnn: Meta-path extracted graph neural network for heterogeneous graph representation learning. Knowledge-Based Systems, 235: 0 107611, 2022. ISSN 0950-7051. doi:https://doi.org/10.1016/j.knosys.2021.107611. URL https://www.sciencedirect.com/science/article/pii/S095070512100873X

    arXiv 2022

Show all 54 references
  1. [9]

    D4explainer: In-distribution explanations of graph neural network via discrete denoising diffusion

    Jialin Chen, Shirley Wu, Abhijit Gupta, and Rex Ying. D4explainer: In-distribution explanations of graph neural network via discrete denoising diffusion. Advances in Neural Information Processing Systems, 36, 2024

    2024

  2. [10]

    How faithful are self-explainable gnns? arXiv preprint arXiv:2308.15096, 2023

    Marc Christiansen, Lea Villadsen, Zhiqiang Zhong, Stefano Teso, and Davide Mottin. How faithful are self-explainable gnns? arXiv preprint arXiv:2308.15096, 2023

    2023 arXiv

  3. [11]

    A density-based algorithm for discovering clusters in large spatial databases with noise

    Martin Ester, Hans-Peter Kriegel, J\" o rg Sander, and Xiaowei Xu. A density-based algorithm for discovering clusters in large spatial databases with noise. In Proceedings of the Second International Conference on Knowledge Discovery and Data Mining, KDD'96, pp.\ 226–231. AAAI...

    1996

  4. [12]

    Meta-path learning for multi-relational graph neural networks

    Francesco Ferrini, Antonio Longa, Andrea Passerini, and Manfred Jaeger. Meta-path learning for multi-relational graph neural networks. In Soledad Villar and Benjamin Chamberlain (eds.), Proceedings of the Second Learning on Graphs Conference, volume 231 of Proceedings of Machi...

    2024

  5. [13]

    Relational deep learning: Graph representation learning on relational databases, 2023

    Matthias Fey, Weihua Hu, Kexin Huang, Jan Eric Lenssen, Rishabh Ranjan, Joshua Robinson, Rex Ying, Jiaxuan You, and Jure Leskovec. Relational deep learning: Graph representation learning on relational databases, 2023. URL https://arxiv.org/abs/2312.04615

    2023 arXiv

  6. [14]

    A review of multi-instance learning assumptions

    James Foulds and Eibe Frank. A review of multi-instance learning assumptions. The knowledge engineering review, 25 0 (1): 0 1--25, 2010

    2010

  7. [15]

    Magnn: Metapath aggregated graph neural network for heterogeneous graph embedding

    Xinyu Fu, Jiani Zhang, Ziqiao Meng, and Irwin King. Magnn: Metapath aggregated graph neural network for heterogeneous graph embedding. In Proceedings of the web conference 2020, pp.\ 2331--2341, 2020

    2020

  8. [16]

    Knowledge transfer for out-of-knowledge-base entities: A graph neural network approach

    Takuo Hamaguchi, Hidekazu Oiwa, Masashi Shimbo, and Yuji Matsumoto. Knowledge transfer for out-of-knowledge-base entities: A graph neural network approach. arXiv preprint arXiv:1706.05674, 2017

    2017 arXiv

  9. [17]

    Heterogeneous graph transformer

    Ziniu Hu, Yuxiao Dong, Kuansan Wang, and Yizhou Sun. Heterogeneous graph transformer. In Proceedings of the web conference 2020, pp.\ 2704--2710, 2020

    2020

  10. [18]

    Johnson, T

    A. Johnson, T. Pollard, O. Badawi, and J. Raffa. eicu collaborative research database demo (version 2.0.1), 2021. URL https://doi.org/10.13026/4mxk-na84

    2021 doi

  11. [19]

    A survey on explainability of graph neural networks

    Jaykumar Kakkad, Jaspal Jannu, Kartik Sharma, Charu Aggarwal, and Sourav Medya. A survey on explainability of graph neural networks. arXiv preprint arXiv:2306.01958, 2023

    2023 arXiv

  12. [20]

    Kipf and Max Welling

    Thomas N. Kipf and Max Welling. Semi-supervised classification with graph convolutional networks. CoRR, abs/1609.02907, 2016. URL http://arxiv.org/abs/1609.02907

    2016 arXiv

  13. [21]

    Graphmse: Efficient meta-path selection in semantically aligned feature space for graph neural networks

    Yi Li, Yilun Jin, Guojie Song, Zihao Zhu, Chuan Shi, and Yiming Wang. Graphmse: Efficient meta-path selection in semantically aligned feature space for graph neural networks. In Proceedings of the AAAI conference on artificial intelligence, volume 35, pp.\ 4206--4214, 2021

    2021

  14. [22]

    Explaining the explainers in graph neural networks: a comparative study

    Antonio Longa, Steve Azzolin, Gabriele Santin, Giulia Cencetti, Pietro Lio, Bruno Lepri, and Andrea Passerini. Explaining the explainers in graph neural networks: a comparative study. ACM Comput. Surv., September 2024. ISSN 0360-0300. doi:10.1145/3696444. URL https://doi.org/1...

    2024 doi

  15. [23]

    Are we really making much progress? revisiting, benchmarking and refining heterogeneous graph neural networks

    Qingsong Lv, Ming Ding, Qiang Liu, Yuxiang Chen, Wenzheng Feng, Siming He, Chang Zhou, Jianguo Jiang, Yuxiao Dong, and Jie Tang. Are we really making much progress? revisiting, benchmarking and refining heterogeneous graph neural networks. In Proceedings of the 27th ACM SIGKDD...

    2021

  16. [24]

    Are we really making much progress? revisiting, benchmarking and refining heterogeneous graph neural networks

    Qingsong Lv, Ming Ding, Qiang Liu, Yuxiang Chen, Wenzheng Feng, Siming He, Chang Zhou, Jianguo Jiang, Yuxiao Dong, and Jie Tang. Are we really making much progress? revisiting, benchmarking and refining heterogeneous graph neural networks. In Proceedings of the 27th ACM SIGKDD...

    2021

  17. [25]

    at Freiburg, Institut f\

    Wolfgang May. Information extraction and integration with Florid : The Mondial case study. Technical Report 131, Universit\"at Freiburg, Institut f\"ur Informatik, 1999. Available from http://dbis.informatik.uni-goettingen.de/Mondial

    1999

  18. [26]

    Interpretable and generalizable graph learning via stochastic attention mechanism

    Siqi Miao, Mia Liu, and Pan Li. Interpretable and generalizable graph learning via stochastic attention mechanism. In International Conference on Machine Learning, pp.\ 15524--15543. PMLR, 2022

    2022

  19. [27]

    Revisiting link prediction on heterogeneous graphs with a multi-view perspective

    Anasua Mitra, Priyesh Vijayan, Sanasam Ranbir Singh, Diganta Goswami, Srinivasan Parthasarathy, and Balaraman Ravindran. Revisiting link prediction on heterogeneous graphs with a multi-view perspective. In 2022 IEEE International Conference on Data Mining (ICDM), pp.\ 358--367...

    2022

  20. [28]

    Prototype-based interpretable graph neural networks

    Alessio Ragno, Biagio La Rosa, and Roberto Capobianco. Prototype-based interpretable graph neural networks. IEEE Transactions on Artificial Intelligence, 5 0 (4): 0 1486--1495, 2022

    2022

  21. [29]

    Lenssen, Yiwen Yuan, Zecheng Zhang, Xinwei He, and Jure Leskovec

    Joshua Robinson, Rishabh Ranjan, Weihua Hu, Kexin Huang, Jiaqi Han, Alejandro Dobles, Matthias Fey, Jan E. Lenssen, Yiwen Yuan, Zecheng Zhang, Xinwei He, and Jure Leskovec. Relbench: A benchmark for deep learning on relational databases, 2024. URL https://arxiv.org/abs/2407.20060

    2024 arXiv

  22. [30]

    Graph networks as learnable physics engines for inference and control, 2018

    Alvaro Sanchez-Gonzalez, Nicolas Heess, Jost Tobias Springenberg, Josh Merel, Martin Riedmiller, Raia Hadsell, and Peter Battaglia. Graph networks as learnable physics engines for inference and control, 2018. URL https://arxiv.org/abs/1806.01242

    2018 arXiv

  23. [31]

    Kipf, Peter Bloem, Rianne van den Berg, Ivan Titov, and Max Welling

    Michael Schlichtkrull, Thomas N. Kipf, Peter Bloem, Rianne van den Berg, Ivan Titov, and Max Welling. Modeling relational data with graph convolutional networks, 2017. URL https://arxiv.org/abs/1703.06103

    2017 arXiv

  24. [32]

    Modeling relational data with graph convolutional networks

    Michael Schlichtkrull, Thomas N Kipf, Peter Bloem, Rianne Van Den Berg, Ivan Titov, and Max Welling. Modeling relational data with graph convolutional networks. In The semantic web: 15th international conference, ESWC 2018, Heraklion, Crete, Greece, June 3--7, 2018, proceeding...

    2018

  25. [33]

    A hierarchy of independence assumptions for multi-relational bayes net classifiers

    Oliver Schulte, Bahareh Bina, Branden Crawford, Derek Bingham, and Yi Xiong. A hierarchy of independence assumptions for multi-relational bayes net classifiers. In 2013 IEEE Symposium on Computational Intelligence and Data Mining (CIDM), pp.\ 150--159. IEEE, 2013

    2013

  26. [34]

    Interpretable prototype-based graph information bottleneck

    Sangwoo Seo, Sungwon Kim, and Chanyoung Park. Interpretable prototype-based graph information bottleneck. In A. Oh, T. Naumann, A. Globerson, K. Saenko, M. Hardt, and S. Levine (eds.), Advances in Neural Information Processing Systems, volume 36, pp.\ 76737--76748. Curran Asso...

    2023

  27. [35]

    Pgm-explainer: Probabilistic graphical model explanations for graph neural networks

    Minh Vu and My T Thai. Pgm-explainer: Probabilistic graphical model explanations for graph neural networks. Advances in neural information processing systems, 33: 0 12225--12235, 2020

    2020

  28. [36]

    Heterogeneous graph attention network

    Xiao Wang, Houye Ji, Chuan Shi, Bai Wang, Yanfang Ye, Peng Cui, and Philip S Yu. Heterogeneous graph attention network. In The world wide web conference, pp.\ 2022--2032, 2019

    2022

  29. [37]

    Gnninterpreter: A probabilistic generative model-level explanation for graph neural networks

    Xiaoqi Wang and Han Wei Shen. Gnninterpreter: A probabilistic generative model-level explanation for graph neural networks. In The Eleventh International Conference on Learning Representations, 2022

    2022

  30. [38]

    Discovering invariant rationales for graph neural networks

    Yingxin Wu, Xiang Wang, An Zhang, Xiangnan He, and Tat-Seng Chua. Discovering invariant rationales for graph neural networks. In International Conference on Learning Representations, 2022

    2022

  31. [39]

    Graph convolutional networks with markov random field reasoning for social spammer detection

    Yongji Wu, Defu Lian, Yiheng Xu, Le Wu, and Enhong Chen. Graph convolutional networks with markov random field reasoning for social spammer detection. In Proceedings of the AAAI conference on artificial intelligence, volume 34, pp.\ 1054--1061, 2020

    2020

  32. [40]

    Gnnexplainer: Generating explanations for graph neural networks

    Zhitao Ying, Dylan Bourgeois, Jiaxuan You, Marinka Zitnik, and Jure Leskovec. Gnnexplainer: Generating explanations for graph neural networks. Advances in neural information processing systems, 32, 2019

    2019

  33. [41]

    Graph information bottleneck for subgraph recognition

    Junchi Yu, Tingyang Xu, Yu Rong, Yatao Bian, Junzhou Huang, and Ran He. Graph information bottleneck for subgraph recognition. In International Conference on Learning Representations, 2020

    2020

  34. [42]

    Heterogeneous graph representation learning with relation awareness

    Le Yu, Leilei Sun, Bowen Du, Chuanren Liu, Weifeng Lv, and Hui Xiong. Heterogeneous graph representation learning with relation awareness. CoRR, abs/2105.11122, 2021. URL https://arxiv.org/abs/2105.11122

    2021 arXiv

  35. [43]

    Heterogeneous graph representation learning with relation awareness

    Le Yu, Leilei Sun, Bowen Du, Chuanren Liu, Weifeng Lv, and Hui Xiong. Heterogeneous graph representation learning with relation awareness. IEEE Transactions on Knowledge and Data Engineering, 35 0 (6): 0 5935--5947, 2022

    2022

  36. [44]

    Xgnn: Towards model-level explanations of graph neural networks

    Hao Yuan, Jiliang Tang, Xia Hu, and Shuiwang Ji. Xgnn: Towards model-level explanations of graph neural networks. In Proceedings of the 26th ACM SIGKDD international conference on knowledge discovery & data mining, pp.\ 430--438, 2020

    2020

  37. [45]

    On explainability of graph neural networks via subgraph explorations

    Hao Yuan, Haiyang Yu, Jie Wang, Kang Li, and Shuiwang Ji. On explainability of graph neural networks via subgraph explorations. In International conference on machine learning, pp.\ 12241--12252. PMLR, 2021

    2021

  38. [46]

    Explainability in graph neural networks: A taxonomic survey

    Hao Yuan, Haiyang Yu, Shurui Gui, and Shuiwang Ji. Explainability in graph neural networks: A taxonomic survey. IEEE transactions on pattern analysis and machine intelligence, 45 0 (5): 0 5782--5799, 2022

    2022

  39. [47]

    Graph transformer networks

    Seongjun Yun, Minbyul Jeong, Raehyun Kim, Jaewoo Kang, and Hyunwoo J Kim. Graph transformer networks. In H. Wallach, H. Larochelle, A. Beygelzimer, F. d Alch\' e -Buc, E. Fox, and R. Garnett (eds.), Advances in Neural Information Processing Systems, volume 32. Curran Associate...

    2019

  40. [48]

    Graph transformer networks

    Seongjun Yun, Minbyul Jeong, Raehyun Kim, Jaewoo Kang, and Hyunwoo J Kim. Graph transformer networks. Advances in neural information processing systems, 32, 2019 b

    2019

  41. [49]

    Graph transformer networks: Learning meta-path graphs to improve gnns

    Seongjun Yun, Minbyul Jeong, Sungdong Yoo, Seunghun Lee, S Yi Sean, Raehyun Kim, Jaewoo Kang, and Hyunwoo J Kim. Graph transformer networks: Learning meta-path graphs to improve gnns. Neural Networks, 153: 0 104--119, 2022 a

    2022

  42. [50]

    Yi, Raehyun Kim, Jaewoo Kang, and Hyunwoo J

    Seongjun Yun, Minbyul Jeong, Sungdong Yoo, Seunghun Lee, Sean S. Yi, Raehyun Kim, Jaewoo Kang, and Hyunwoo J. Kim. Graph transformer networks: Learning meta-path graphs to improve gnns. Neural Networks, 153: 0 104--119, 2022 b . ISSN 0893-6080. doi:https://doi.org/10.1016/j.ne...

    2022 doi

  43. [51]

    Protgnn: Towards self-explaining graph neural networks

    Zaixi Zhang, Qi Liu, Hao Wang, Chengqiang Lu, and Cheekong Lee. Protgnn: Towards self-explaining graph neural networks. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 36, pp.\ 9127--9135, 2022

    2022

  44. [52]

    Towards robust fidelity for evaluating explainability of graph neural networks, 2023

    Xu Zheng, Farhad Shirani, Tianchun Wang, Wei Cheng, Zhuomin Chen, Haifeng Chen, Hua Wei, and Dongsheng Luo. Towards robust fidelity for evaluating explainability of graph neural networks, 2023

    2023

  45. [53]

    Relation structure-aware heterogeneous graph neural network

    Shichao Zhu, Chuan Zhou, Shirui Pan, Xingquan Zhu, and Bin Wang. Relation structure-aware heterogeneous graph neural network. In 2019 IEEE international conference on data mining (ICDM), pp.\ 1534--1539. IEEE, 2019

    2019

  46. [54]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 gl...

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.