Triadic Closure-Heterogeneity-Harmony GCN for Link Prediction

Junfan Yi; Ke-ke Shang; Michael Small; Yijie Zhou

REVIEW 4 major objections 5 minor 59 references

Triadic Closure-Heterogeneity-Harmony GCN for Link Prediction

T0 review · 4 major / 5 minor · reviewed 2026-08-16 · deepseek-v4-flash

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

Pith's one-line read TriHetGCN claims that weighting GCN message passing by common-neighbor counts and degree differences achieves higher link prediction accuracy than 10 baselines on all nine tested networks.

desk verdict The TriHetGCN architecture is a plausible and tidy extension of GCN, but the paper's SOTA claim rests on an evaluation that likely leaks test edges and lacks variance bars. read the letter →

arxiv 2504.20492 v1 pith:QN7LJTMN submitted 2025-04-29 cs.SI physics.data-anphysics.soc-ph

Ke-ke Shang , Junfan Yi , Michael Small , Yijie Zhou This is my paper · ORCID

classification cs.SIphysics.data-anphysics.soc-ph

keywords linkpredictiongraphconvolutionalnetworkstriadicclosuredegreeheterogeneitycommonneighborsnodeattributescomplexanchornodes

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 tries to show that a graph convolutional network can be made more accurate for link prediction by explicitly weighting its aggregation with two classic network-physics cues: how many common neighbors a node pair shares and how different their degrees are. It also tries to show that networks without node attributes can still be fed into a GCN by using normalized shortest-path distances to high-degree anchor nodes as synthetic node features. If true, this matters because it connects statistical-physics heuristics to deep graph learning, letting one model handle both attributed and purely topological networks. On nine datasets the model reports higher AUC and AP than heuristic and GNN baselines, with the largest gains on sparse, featureless networks.

What carries the argument

The carrying object is the topology-aware propagation operator $\Phi$ with entries $\varphi_{ij} = D^{-1/2}_{ii} D^{-1/2}_{jj} \cdot \exp(s_{cn}\mathrm{CN}_{ij} + s_{hi}\mathrm{HI}_{ij})$, where $\mathrm{CN}_{ij}$ is the number of common neighbors of nodes $i$ and $j$, $\mathrm{HI}_{ij} = |d_i - d_j|$ is the absolute degree difference, and $s_{cn}, s_{hi}$ are learnable scalars. This operator replaces the normalized adjacency matrix in a GCN, up-weighting message flow between node pairs that are triangle-prone or degree-heterogeneous. The second mechanism is pseudo-attribute construction: for featureless graphs, each node's feature vector is the vector of normalized shortest-path distances to selected high-degree anchor nodes, which lets the GCN perceive node position and global structure.

What would settle it

Rerun every method on the full Twitter graph (or all methods on the same 4-hop subgraph), repeat over at least 10 random splits, and check whether TriHetGCN's AUC margin over GCN, which ranges from about 0.02 to 0.96 percentage points on the reported tables, and its margin over heuristics survive with confidence intervals.

Watch

Extended reading notes

Core claim

The central claim is that link prediction accuracy improves if triadic closure and degree heterogeneity are placed inside the GCN aggregation rather than used only as separate scoring heuristics. The paper defines a propagation operator in which each edge's message-passing weight is the standard GCN normalization multiplied by an exponential of a learnable-weighted combination of the pair's common-neighbor count and absolute degree difference. For graphs with no node attributes, node features are built from normalized shortest-path distances to a set of high-degree anchor nodes, giving the GCN global position information. Candidate links are then scored by the Hadamard product of learned node representations passed through a two-layer fully connected network, trained with binary cross-entropy. The paper reports that this model outperforms common-neighbor, Adamic-Adar, resource-allocation, Katz, random-walk, GCN, GraphSAGE, and GAT baselines on AUC and AP across Cora, Citeseer, PubMed, DBLP, CS, Facebook, Power, Twitter, and INT.

Load-bearing premise

The central claim depends on the evaluation being fair and the reported margins being real: results come from a single fixed train, validation, and test split with seed 42, negative edges are randomly sampled, no variance or significance tests are reported, and on Twitter the heuristic baselines run on a 4-hop subgraph while the GNNs run on the full graph.

Editorial extensions

If this is right

A standard GCN can be upgraded for link prediction simply by re-weighting its propagation with common-neighbor counts and degree differences, without changing the decoder or the loss function.
Networks without node attributes become tractable for GCNs through anchor-distance pseudo-features, so the same model handles attributed citation and coauthor graphs as well as purely topological power and internet graphs.
The learnable weights on the two indicators allow the model to adapt to each network's structure, favoring triadic closure where clustering is high and degree heterogeneity where hubs dominate.
The reported margins over GCN are modest on attributed networks but larger on featureless ones, suggesting the biggest practical payoff is for large sparse graphs that lack node attributes.

Reading between the lines

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

Beyond the paper's own claims, the same pair-level weighting mechanism could be dropped into other message-passing architectures, since it is computed as a scalar per edge before aggregation.
The anchor-distance feature scheme is effectively a global positional encoding; a natural test the paper does not run is whether anchor count and anchor choice (highest degree versus random or spectral) change the reported margins.
The reported Twitter comparison evaluates heuristic baselines on a 4-hop subgraph while GNNs use the full graph, so a full-graph run of Katz or local random walk is a direct way to test whether the claimed advantage over global heuristics is real.
The ablation pattern suggests a practical decision rule—common-neighbor weighting on high-clustering networks and degree-difference weighting on hub-dominated networks—that the paper documents but does not formalize.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Desk Editor's Note

read the letter

You should know two things about arXiv:2504.20492. First, the core idea is real: modulating GCN message-passing weights with common-neighbor counts and degree differences, plus anchor-distance pseudo-features for featureless networks, is a coherent and interpretable way to inject triadic closure and heterogeneity into graph learning. The ablation study behaves as you'd expect. Second, the empirical claim that it beats everything on nine datasets is not yet supported, because the method as written computes structural features on the full graph.

The leakage worry is the load-bearing problem. Algorithm 1 takes the full G=(V,E), finds connected components, picks anchors, and computes shortest paths. Section 3.5 defines CN and HI for node pairs without ever saying that validation/test edges were removed first. The paper only says the dataset is split for training/validation/test, but never states that the pseudo-features and the phi_ij weights are computed on the training subgraph. If they are computed on the full graph, test edges directly shape both the node features and the aggregation weights, and the small margins over GCN (e.g., DBLP 96.34 vs 96.30, PubMed 97.21 vs 97.18) are exactly what leakage would produce. The authors may have done it correctly, but the manuscript doesn't say so, and that omission matters.

The secondary problems are statistical and comparative. No error bars anywhere; the footnote mentions 10 repeated runs for GNN methods but reports no variance or significance. The Twitter comparison evaluates heuristics on a 4-hop subgraph while GNNs use the full graph; the footnote discloses this, but it doesn't make the comparison fair. The baseline set is thin—GCN, GraphSAGE, GAT—and omits standard strong link predictors like SEAL or GAE, so "SOTA" is against a weak field. Code is promised but not yet public.

None of this kills the idea. The architecture is simple, sensible, and consistent with the interpretability goals. A reader interested in physics-informed GNN variants will find it useful. But I would not cite the empirical SOTA claim until the leakage question is answered and variance appears.

Recommendation: don't desk reject. Send to peer review with the expectation that referees verify the training-graph-only computation, demand standard deviations, and ask for at least one or two stronger baselines. The core idea deserves a fair shot; the current evidence doesn't.

Referee Report

4 major / 5 minor

Summary. The paper proposes TriHetGCN, a GCN extension for link prediction that injects two explicit topological indicators, common-neighbor count and degree-difference heterogeneity, into the message-passing operator through a learned exponential weighting, and that constructs pseudo node features for attribute-free networks from shortest-path distances to high-degree anchor nodes. The model is evaluated on nine datasets, comparing against classical heuristics and three GNN baselines, with the central claim that it consistently outperforms all baselines on both AUC and AP. The manuscript also includes an ablation separating the CN and HI contributions.

Significance. If the reported results were validated, the contribution would be a useful and interpretable bridge between statistical-physics heuristics and GNN-based link prediction, with a concrete mechanism for handling attribute-free networks and an ablation that connects model components to network structure. The paper's strengths are its broad evaluation across feature-rich and featureless graphs, the explicit anchoring of the architecture in triadic closure and degree heterogeneity, and the clear statement of the proposed topology-feature construction. However, the empirical claims are not yet established because of leakage risk in the structural feature computation, an unfair Twitter comparison, the absence of variance reporting and significance tests, and a baseline set that is too narrow for the stated state-of-the-art claim. The code is also not currently available, so the reported numbers cannot be independently reproduced.

major comments (4)

[Section 3.4, Algorithm 1; Section 3.5] Algorithm 1 takes the full graph G=(V,E) as input and computes connected components, anchor selection, and shortest-path distances, and Section 3.5 defines CN and HI without excluding validation/test edges. Under the split described in Section 3.1, these quantities must be computed from the training graph only; otherwise test edges directly influence the anchor features, CN/HI weights, and the message-passing operator in Eq. (8). The paper does not state that this is done. Please clarify the implementation and rerun the experiments with all structural features and anchor selection computed on the training subgraph; if results change materially, the state-of-the-art claim in Section 4.3 would be an artifact of leakage.
[Section 4.3, Tables 4 and 5] All results are reported without variance or significance tests, and some differences are very small (PubMed AUC 97.21 vs 97.18; DBLP 96.34 vs 96.30; Facebook 99.52 vs 99.50). The Table 4 footnote states that heuristics were repeated 100 times and GNN-based methods 10 times, but it also states that a fixed seed (42) is used for the split; with a fixed seed, repeated runs are identical and cannot estimate sampling variability. Report means and standard deviations over multiple random splits, and include a paired significance test for the TriHetGCN-versus-GCN comparison.
[Section 4.3, Table 4 footnote] The Twitter row is not a controlled comparison: Katz, RWR, LP, and LRW are evaluated on a 4-hop subgraph with 3,303 nodes and 4,711 edges, while GCN, GraphSAGE, GAT, and TriHetGCN are evaluated on the full graph with 256,491 nodes and 327,374 edges. The large reported gaps (e.g., 54.04 vs 94.21 AUC) therefore cannot be attributed to model quality. Evaluate all methods on the same graph, or present the heuristic results on the full graph where feasible.
[Section 4.3, Abstract] The 'state-of-the-art' claim is stronger than the baseline set supports. The only learning-based baselines are GCN, GraphSAGE, and GAT; recent strong link-prediction GNNs (e.g., SEAL, NCN, Neo-GNN, BUDDY) are not compared. Either include representative recent methods or restrict the claim to the compared baselines.

minor comments (5)

[Section 4.1] The text says 'three synthetic graph datasets,' but Power, Twitter, and INT are real-world datasets; this appears to be a wording error, as these datasets are featureless rather than synthetic.
[Section 3.3, Section 4.3] There are typographical errors, including 'GGNs' instead of 'GNNs' and 'prob sets' instead of 'test sets'; the manuscript should be proofread.
[Section 4.2, Algorithm 1] The maximum number of anchors per component, m, is left blank in the text ('at most .'), and the minimum component size smin is not defined; without these values the pseudo-feature construction is not fully reproducible.
[Equation (8)] The diagonal entries of Phi for i=j are not defined; if self-loops are added to the adjacency matrix for GCN-style propagation, please specify CN_ii and HI_ii (or set them to zero).
[Data Availability] The repository is not public ('to be made public upon acceptance'); please provide code and a configuration file during review, or state the exact implementation details for all hyperparameters including the anchor cap.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: TriHetGCN's SOTA claim is an empirical result benchmarked against external baselines; no prediction reduces to fitted inputs or to self-citation by construction.

full rationale

TriHetGCN is an empirical method paper; its central claim is the consistent SOTA comparison in Sec. 4.3, supported by AUC/AP tables against heuristic and GNN baselines. The methodological components—anchor-distance pseudo-features (Eq. 5), CN (Eq. 6), HI (Eq. 7), and exponential structural weighting (Eq. 8)—are design choices introduced and then tested, not quantities derived from the target result. The heterogeneity index is attributed to the authors' own prior work [2], but that reference supplies a heuristic definition (degree difference), not a uniqueness theorem, a fitted parameter, or an argument forbidding alternatives; the paper's validity does not rest on that citation being an externally established law. The ablation study (Sec. 4.4) independently tests each injected heuristic against the vanilla GCN baseline on the same splits. Therefore no equation or fitted value is renamed as a prediction, and no load-bearing premise is justified solely by self-citation. Possible concerns about computing features on the full graph and single-seed evaluation are evaluation-validity issues, not circularity: the claimed margins are not equal to the model's inputs by construction.

Assumptions & free parameters 6 free parameters · 5 assumptions · 0 invented entities

The central claim is empirical, and it rests on standard GNN machinery plus several modeling choices and evaluation assumptions. No new physical entities are introduced.

free parameters (6)

s_cn = learned
Learnable scalar controlling the contribution of common neighbors in Eq. (8).
s_hi = learned
Learnable scalar controlling the contribution of degree heterogeneity in Eq. (8).
anchor selection ratio r = 0.15
Fraction of highest-degree nodes selected as anchors per component in Algorithm 1.
max anchors per component m = 150
Cap on the number of anchor nodes selected from any single component in Algorithm 1.
component size threshold = 80th percentile
Components whose cumulative size accounts for at least 80% of nodes are used for anchor selection.
initialization range for s_cn and s_hi = [0, 0.5]
Both learnable scalars are initialized randomly in this range, as stated in Section 4.2.

assumptions (5)

domain assumption The graph is undirected and static, with no multiple links or self-links.
Stated in Section 3.1 as part of the problem formulation.
domain assumption Negative samples are randomly selected non-linked node pairs, equal in number to positive links.
Described in Section 3.1 and implemented via PyG's RandomLinkSplit.
domain assumption The universal set U of all node pairs and the assumption that test links are missing uniformly.
Introduced in Section 3.1 as the basis for link prediction evaluation.
domain assumption Common neighbor count and degree difference are informative topological indicators for link prediction.
This is the central modeling hypothesis of the paper, motivating Eq. (8).
standard math Standard GCN message passing and backpropagation are assumed as background knowledge.
The method builds on the GCN formulation from Kipf and Welling [30].

how reviews work

0 comments

Cite this review

Pith. "Pith review of Triadic Closure-Heterogeneity-Harmony GCN for Link Prediction." pith.science (2026). https://pith.science/paper/QN7LJTMN

@misc{pith2026250420492,
  author       = {Pith},
  title        = {Pith review of: Triadic Closure-Heterogeneity-Harmony GCN for Link Prediction},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/QN7LJTMN}},
  note         = {Machine review of arXiv:2504.20492}
}

read the original abstract

Link prediction aims to estimate the likelihood of connections between pairs of nodes in complex networks, which is beneficial to many applications from friend recommendation to metabolic network reconstruction. Traditional heuristic-based methodologies in the field of complex networks typically depend on predefined assumptions about node connectivity, limiting their generalizability across diverse networks. While recent graph neural network (GNN) approaches capture global structural features effectively, they often neglect node attributes and intrinsic structural relationships between node pairs. To address this, we propose TriHetGCN, an extension of traditional Graph Convolutional Networks (GCNs) that incorporates explicit topological indicators -- triadic closure and degree heterogeneity. TriHetGCN consists of three modules: topology feature construction, graph structural representation, and connection probability prediction. The topology feature module constructs node features using shortest path distances to anchor nodes, enhancing global structure perception. The graph structural module integrates topological indicators into the GCN framework to model triadic closure and heterogeneity. The connection probability module uses deep learning to predict links. Evaluated on nine real-world datasets, from traditional networks without node attributes to large-scale networks with rich features, TriHetGCN achieves state-of-the-art performance, outperforming mainstream methods. This highlights its strong generalization across diverse network types, offering a promising framework that bridges statistical physics and graph deep learning.

Figures

Figures reproduced from arXiv: 2504.20492 by the authors.

**Figure 2.** An overview of the proposed TriHetGCN framework, which integrates physical statistic rules and GCN [PITH_FULL_IMAGE:figures/full_fig_p007_2.png] view at source ↗

**Figure 3.** Network Heterogeneity and Triangle Structure: The distribution of these two structural properties across [PITH_FULL_IMAGE:figures/full_fig_p013_3.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

59 extracted references · 42 canonical work pages

[1]

The link prediction problem for social networks

David Liben-Nowell and Jon Kleinberg. The link prediction problem for social networks. In Proceedings of the twelfth international conference on Information and knowledge management, pages 556–559, 2003

2003
[2]

Link prediction for tree-like networks

Ke-ke Shang, Tong-chen Li, Michael Small, David Burton, and Yan Wang. Link prediction for tree-like networks. Chaos: An Interdisciplinary Journal of Nonlinear Science, 29(6), 2019

work page 2019
[3]

Predicting future links with new nodes in temporal academic networks

Yijun Ran, Si-Yuan Liu, Xiaoyao Yu, Ke-Ke Shang, and Tao Jia. Predicting future links with new nodes in temporal academic networks. Journal of Physics: Complexity, 3(1):015006, 2022

work page 2022
[4]

A supervised machine learning link predic- tion approach for academic collaboration recommendation

Nesserine Benchettara, Rushed Kanawati, and Céline Rouveirol. A supervised machine learning link predic- tion approach for academic collaboration recommendation. In Proceedings of the fourth ACM conference on Recommender systems, pages 253–256, 2010

work page 2010
[5]

Boostgapfill: improving the fidelity of metabolic network reconstructions through integrated constraint and pattern-based methods.Bioin- formatics, 33(4):608–611, 2017

Tolutola Oyetunde, Muhan Zhang, Yixin Chen, Yinjie Tang, and Cynthia Lo. Boostgapfill: improving the fidelity of metabolic network reconstructions through integrated constraint and pattern-based methods.Bioin- formatics, 33(4):608–611, 2017

work page 2017
[6]

Novel drug- target interactions via link prediction and network embedding

E Amiri Souri, Roman Laddach, SN Karagiannis, Lazaros G Papageorgiou, and Sophia Tsoka. Novel drug- target interactions via link prediction and network embedding. BMC bioinformatics, 23(1):121, 2022

work page 2022
[7]

Recommendation as link prediction in bipartite graphs: A graph kernel-based machine learning approach

Xin Li and Hsinchun Chen. Recommendation as link prediction in bipartite graphs: A graph kernel-based machine learning approach. Decision Support Systems, 54(2):880–890, 2013

work page 2013
[8]

A review of relational machine learning for knowledge graphs

Maximilian Nickel, Kevin Murphy, V olker Tresp, and Evgeniy Gabrilovich. A review of relational machine learning for knowledge graphs. Proceedings of the IEEE, 104(1):11–33, 2015

2015

Show all 59 references

[9]

Link prediction in complex networks: A survey

Linyuan Lü and Tao Zhou. Link prediction in complex networks: A survey. Physica A: Statistical Mechanics and its Applications, 390(6):1150–1170, 2011

2011
[10]

Toward link predictability of complex networks

Linyuan Lü, Liming Pan, Tao Zhou, Yi-Cheng Zhang, and H Eugene Stanley. Toward link predictability of complex networks. Proceedings of the National Academy of Sciences, 112(8):2325–2330, 2015

2015
[11]

Link prediction for long-circle-like networks

Ke-ke Shang and Michael Small. Link prediction for long-circle-like networks. Physical review E , 105(2):024311, 2022

2022
[12]

Link prediction in weighted networks: The role of weak ties

Linyuan Lü and Tao Zhou. Link prediction in weighted networks: The role of weak ties. EPL (Europhysics Letters), 89(1):18001, 2010

2010
[13]

The degree- related clustering coefficient and its application to link prediction

Yangyang Liu, Chengli Zhao, Xiaojie Wang, Qiangjuan Huang, Xue Zhang, and Dongyun Yi. The degree- related clustering coefficient and its application to link prediction. Physica A: Statistical Mechanics and its Applications, 454:24–33, 2016

2016
[14]

Finding missing edges in networks based on their community structure

Bowen Yan and Steve Gregory. Finding missing edges in networks based on their community structure. Physical Review E, 85(5):056112, 2012

2012
[15]

A novel metric for community detection

Ke-ke Shang, Michael Small, Yan Wang, Di Yin, and Shu Li. A novel metric for community detection. EPL (Europhysics Letters), 129(6):68002, 2020

2020
[16]

Machine learning informed by micro and mesoscopic statistical physics methods for community detection

Yijun Ran, Junfan Yi, Wei Si, Michael Small, and Ke-ke Shang. Machine learning informed by micro and mesoscopic statistical physics methods for community detection. arXiv preprint arXiv:2504.13538, 2025

2025 arXiv
[17]

A new status index derived from sociometric analysis

Leo Katz. A new status index derived from sociometric analysis. Psychometrika, 18(1):39–43, 1953

1953
[18]

Resistance distance

Douglas J Klein and Milan Randi ´c. Resistance distance. Journal of Mathematical Chemistry, 12(1):81–95, 1993

1993
[19]

Fouss, A

F. Fouss, A. Pirotte, J. Renders, and M. Saerens. Random-walk computation of similarities between nodes of a graph with application to collaborative recommendation. IEEE Transactions on Knowledge and Data Engineering, 19(3):355–369, 2007. 14

2007
[20]

Learning structural features of nodes in large-scale networks for link prediction

Aakas Zhiyuli, Xun Liang, and Xiaoping Zhou. Learning structural features of nodes in large-scale networks for link prediction. In Proceedings of the Thirtieth AAAI Conference on Artificial Intelligence , AAAI’16, pages 4286–4287, Phoenix, Arizona, USA, Feb. 2016. AAAI Press

2016
[21]

Deep learning based link prediction with social pattern and external attribute knowledge in bibliographic networks

Chuanting Zhang, Haixia Zhang, Dongfeng Yuan, and Minggao Zhang. Deep learning based link prediction with social pattern and external attribute knowledge in bibliographic networks. In 2016 IEEE International Conference on Smart Data (SmartData), pages 815–821, 2016

2016
[22]

Adaptive similarity function with structural features of network embedding for missing link prediction

Chuanting Zhang, Ke-Ke Shang, and Jingping Qiao. Adaptive similarity function with structural features of network embedding for missing link prediction. Complexity, 2021(1):1277579, 2021

2021
[23]

A comprehensive survey on graph neural networks

Zonghan Wu, Shirui Pan, Fengwen Chen, Guodong Long, Chengqi Zhang, and Philip S Yu. A comprehensive survey on graph neural networks. IEEE transactions on neural networks and learning systems , 32(1):4–24, 2020

2020
[24]

Link prediction using supervised learning

Mohammad Al Hasan, Vineet Chaoji, Saeed Salem, and Mohammed Zaki. Link prediction using supervised learning. In SDM06: workshop on link analysis, counter-terrorism and security, volume 30, pages 798–805, 2006

2006
[25]

Compound rank- k projections for bilinear analysis

Xiaojun Chang, Feiping Nie, Sen Wang, Yi Yang, Xiaofang Zhou, and Chengqi Zhang. Compound rank- k projections for bilinear analysis. IEEE Transactions on Neural Networks and Learning Systems, 27(7):1502– 1513, 2016

2016
[26]

Self-supervised deep correlation track- ing

Di Yuan, Xiaojun Chang, Po-Yao Huang, Qiao Liu, and Zhenyu He. Self-supervised deep correlation track- ing. IEEE Transactions on Image Processing, 30:976–985, 2021

2021
[27]

The graph neural network model

Franco Scarselli, Marco Gori, Ah Chung Tsoi, Markus Hagenbuchner, and Gabriele Monfardini. The graph neural network model. IEEE transactions on neural networks, 20(1):61–80, 2008

2008
[28]

Relational inductive biases, deep learning, and graph networks

Peter W Battaglia, Jessica B Hamrick, Victor Bapst, Alvaro Sanchez-Gonzalez, Vinicius Zambaldi, Mateusz Malinowski, Andrea Tacchetti, David Raposo, Adam Santoro, Ryan Faulkner, et al. Relational inductive biases, deep learning, and graph networks. arXiv preprint arXiv:1806.01261, 2018

2018 arXiv
[29]

Line graph neural networks for link prediction

Lei Cai, Jundong Li, Jie Wang, and Shuiwang Ji. Line graph neural networks for link prediction. IEEE Transactions on Pattern Analysis and Machine Intelligence, pages 1–1, 2021

2021
[30]

Semi-supervised classification with graph convolutional networks

Thomas N Kipf and Max Welling. Semi-supervised classification with graph convolutional networks. arXiv preprint arXiv:1609.02907, 2016

2016 arXiv
[31]

Gcn-gan: A non-linear temporal link prediction model for weighted dynamic networks

Kai Lei, Meng Qin, Bo Bai, Gong Zhang, and Min Yang. Gcn-gan: A non-linear temporal link prediction model for weighted dynamic networks. In IEEE INFOCOM 2019-IEEE conference on computer communi- cations, pages 388–396. IEEE, 2019

2019
[32]

Temporal link prediction in directed networks based on self-attention mechanism

Jinsong Li, Jianhua Peng, Shuxin Liu, Lintianran Weng, and Cong Li. Temporal link prediction in directed networks based on self-attention mechanism. Intelligent Data Analysis, 26(1):173–188, 2022

2022
[33]

Link prediction model for opportunistic networks based on feature fusion

Jian Shu, Jiawei Shi, and Liang Liao. Link prediction model for opportunistic networks based on feature fusion. IEEE Access, 10:80900–80909, 2022

2022
[34]

Friends and neighbors on the web

Lada A Adamic and Eytan Adar. Friends and neighbors on the web. Social networks, 25(3):211–230, 2003

2003
[35]

Clustering and preferential attachment in growing networks

Mark EJ Newman. Clustering and preferential attachment in growing networks. Physical review E , 64(2):025102, 2001

2001
[36]

Emergence of scaling in random networks

Albert-László Barabási and Réka Albert. Emergence of scaling in random networks. Science, 286(5439):509–512, 1999

1999
[37]

Predicting missing links via local information

Tao Zhou, Linyuan Lü, and Yi-Cheng Zhang. Predicting missing links via local information. The European Physical Journal B, 71:623–630, 2009

2009
[38]

Link prediction with node clustering coefficient

Zhihao Wu, Youfang Lin, Jing Wang, and Steve Gregory. Link prediction with node clustering coefficient. Physica A: Statistical Mechanics and its Applications, 452:1–8, 2016. 15

2016
[39]

The role of direct links for link prediction in evolving networks

Ke Ke Shang, Michael Small, Xiao Ke Xu, and Wei Sheng Yan. The role of direct links for link prediction in evolving networks. EPL, 117(2):28002, 2017

2017
[40]

Ensemble-model-based link prediction of complex networks

Kuanyang Li, Lilan Tu, and Lang Chai. Ensemble-model-based link prediction of complex networks. Com- puter Networks, 166:106978, 2020

2020
[41]

Graph learning based recommender systems: A review

Shoujin Wang, Liang Hu, Yan Wang, Xiangnan He, Quan Z Sheng, Mehmet A Orgun, Longbing Cao, Francesco Ricci, and Philip S Yu. Graph learning based recommender systems: A review. arXiv preprint arXiv:2105.06339, 2021

2021 arXiv
[42]

Graph attention networks

Petar Velickovic, Guillem Cucurull, Arantxa Casanova, Adriana Romero, Pietro Lio, Yoshua Bengio, et al. Graph attention networks. stat, 1050(20):10–48550, 2017

2017
[43]

Inductive representation learning on large graphs

Will Hamilton, Zhitao Ying, and Jure Leskovec. Inductive representation learning on large graphs. Advances in neural information processing systems, 30, 2017

2017
[44]

Gc-lstm: Graph convolution embedded lstm for dynamic network link prediction

Jinyin Chen, Xueke Wang, and Xuanheng Xu. Gc-lstm: Graph convolution embedded lstm for dynamic network link prediction. Applied Intelligence, pages 1–16, 2022

2022
[45]

The meaning and use of the area under a receiver operating charac- teristic (roc) curve

James A Hanley and Barbara J McNeil. The meaning and use of the area under a receiver operating charac- teristic (roc) curve. Radiology, 143(1), 1982

1982
[46]

A statistical method for system evaluation using incom- plete judgments

Javed A Aslam, Virgil Pavlu, and Emine Yilmaz. A statistical method for system evaluation using incom- plete judgments. In Proceedings of the 29th annual international ACM SIGIR conference on Research and development in information retrieval, pages 541–548, 2006

2006
[47]

A new interpretation of average precision

Stephen Robertson. A new interpretation of average precision. InProceedings of the 31st annual international ACM SIGIR conference on Research and development in information retrieval, pages 689–690, 2008

2008
[48]

Scikit-learn: Machine learning in python

Fabian Pedregosa, Gaël Varoquaux, Alexandre Gramfort, Vincent Michel, Bertrand Thirion, Olivier Grisel, Mathieu Blondel, Peter Prettenhofer, Ron Weiss, Vincent Dubourg, et al. Scikit-learn: Machine learning in python. the Journal of machine Learning research, 12:2825–2830, 2011

2011
[49]

Collec- tive classification in network data

Prithviraj Sen, Galileo Namata, Mustafa Bilgic, Lise Getoor, Brian Galligher, and Tina Eliassi-Rad. Collec- tive classification in network data. AI magazine, 29(3):93–93, 2008

2008
[50]

Revisiting semi-supervised learning with graph embeddings

Zhilin Yang, William Cohen, and Ruslan Salakhudinov. Revisiting semi-supervised learning with graph embeddings. In International conference on machine learning, pages 40–48. PMLR, 2016

2016
[51]

Deep gaussian embedding of graphs: Unsupervised induc- tive learning via ranking

Aleksandar Bojchevski and Stephan Günnemann. Deep gaussian embedding of graphs: Unsupervised induc- tive learning via ranking. arXiv preprint arXiv:1707.03815, 2017

2017 arXiv
[52]

Pitfalls of graph neural network evaluation

Oleksandr Shchur, Maximilian Mumme, Aleksandar Bojchevski, and Stephan Günnemann. Pitfalls of graph neural network evaluation. arXiv preprint arXiv:1811.05868, 2018

2018 arXiv
[53]

Learning to discover social circles in ego networks

Jure Leskovec and Julian Mcauley. Learning to discover social circles in ego networks. Advances in neural information processing systems, 25, 2012

2012
[54]

Collective dynamics of ‘small-world’networks

Duncan J Watts and Steven H Strogatz. Collective dynamics of ‘small-world’networks. nature, 393(6684):440–442, 1998

1998
[55]

The anatomy of a scientific rumor

Manlio De Domenico, Antonio Lima, Paul Mougel, and Mirco Musolesi. The anatomy of a scientific rumor. Scientific reports, 3(1):2980, 2013

2013
[56]

Measuring isp topologies with rocket- fuel

Neil Spring, Ratul Mahajan, David Wetherall, and Thomas Anderson. Measuring isp topologies with rocket- fuel. IEEE/ACM Transactions on networking, 12(1):2–16, 2004

2004
[57]

Similarity index based on local paths for link prediction of complex networks

Linyuan Lü, Ci-Hang Jin, and Tao Zhou. Similarity index based on local paths for link prediction of complex networks. Physical Review E—Statistical, Nonlinear, and Soft Matter Physics, 80(4):046122, 2009

2009
[58]

The anatomy of a large-scale hypertextual web search engine

Sergey Brin and Lawrence Page. The anatomy of a large-scale hypertextual web search engine. Computer networks and ISDN systems, 30(1-7):107–117, 1998

1998
[59]

Link prediction based on local random walk.Europhysics Letters, 89(5):58007, 2010

Weiping Liu and Linyuan Lü. Link prediction based on local random walk.Europhysics Letters, 89(5):58007, 2010. 16

2010

Pith tools

Reviewed August 16, 2026 · model on record in the stance chip above.

[1] [1]

The link prediction problem for social networks

David Liben-Nowell and Jon Kleinberg. The link prediction problem for social networks. In Proceedings of the twelfth international conference on Information and knowledge management, pages 556–559, 2003

2003

[2] [2]

Link prediction for tree-like networks

Ke-ke Shang, Tong-chen Li, Michael Small, David Burton, and Yan Wang. Link prediction for tree-like networks. Chaos: An Interdisciplinary Journal of Nonlinear Science, 29(6), 2019

work page 2019

[3] [3]

Predicting future links with new nodes in temporal academic networks

Yijun Ran, Si-Yuan Liu, Xiaoyao Yu, Ke-Ke Shang, and Tao Jia. Predicting future links with new nodes in temporal academic networks. Journal of Physics: Complexity, 3(1):015006, 2022

work page 2022

[4] [4]

A supervised machine learning link predic- tion approach for academic collaboration recommendation

Nesserine Benchettara, Rushed Kanawati, and Céline Rouveirol. A supervised machine learning link predic- tion approach for academic collaboration recommendation. In Proceedings of the fourth ACM conference on Recommender systems, pages 253–256, 2010

work page 2010

[5] [5]

Boostgapfill: improving the fidelity of metabolic network reconstructions through integrated constraint and pattern-based methods.Bioin- formatics, 33(4):608–611, 2017

Tolutola Oyetunde, Muhan Zhang, Yixin Chen, Yinjie Tang, and Cynthia Lo. Boostgapfill: improving the fidelity of metabolic network reconstructions through integrated constraint and pattern-based methods.Bioin- formatics, 33(4):608–611, 2017

work page 2017

[6] [6]

Novel drug- target interactions via link prediction and network embedding

E Amiri Souri, Roman Laddach, SN Karagiannis, Lazaros G Papageorgiou, and Sophia Tsoka. Novel drug- target interactions via link prediction and network embedding. BMC bioinformatics, 23(1):121, 2022

work page 2022

[7] [7]

Recommendation as link prediction in bipartite graphs: A graph kernel-based machine learning approach

Xin Li and Hsinchun Chen. Recommendation as link prediction in bipartite graphs: A graph kernel-based machine learning approach. Decision Support Systems, 54(2):880–890, 2013

work page 2013

[8] [8]

A review of relational machine learning for knowledge graphs

Maximilian Nickel, Kevin Murphy, V olker Tresp, and Evgeniy Gabrilovich. A review of relational machine learning for knowledge graphs. Proceedings of the IEEE, 104(1):11–33, 2015

2015