Pith. sign in

REVIEW 5 major objections 7 minor 1 cited by

Machine Learning Informed by Micro and Mesoscopic Statistical Physics Methods for Community Detection

T0 review · 5 major / 7 minor · reviewed 2026-08-16 · deepseek-v4-flash

Pith's one-line read The paper claims that a machine-learning layer that weights node pairs by predicted same-community likelihood improves community detection on both synthetic and real networks, and that the largest accuracy gains appear when the layer is…

desk verdict Useful self-training idea, but the headline NMI/ARI gains come from training on the same labels used for scoring—the stress-test concern holds. read the letter →

arxiv 2504.13538 v1 pith:KE5CCSQV submitted 2025-04-18 cs.SI nlin.AOphysics.soc-ph

classification cs.SInlin.AOphysics.soc-ph
keywords communitydetectionnode-pairsimilarityensemblelearningstatisticalphysicsmodularitynormalizedmutualinformationadjustedrandindexweightednetwork
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper tries to show that community detection improves when node-pair similarity is treated as a learnable signal and injected into standard detection algorithms. The proposed framework first obtains a preliminary partition of the network, or uses known community labels when those are available, then trains an ensemble of decision trees, random forests, and XGBoost to predict whether two nodes belong to the same community from three micro-level features. The predicted likelihoods become edge weights in a weighted similarity network, and four classic algorithms (Louvain, Leiden, Infomap, and Fast-Greedy) are re-run on that re-weighted graph. Across one artificial and five real-world networks, the paper reports higher modularity, normalized mutual information, and adjusted rand index than the same algorithms achieve on the original graph. The paper also reports a strong correlation between the internal-versus-external node-pair similarity gap and detection accuracy, and offers that correlation as the mechanism behind the gains.

What carries the argument

The load-bearing mechanism is the weighted similarity network: node-pair likelihoods predicted by an ensemble of Decision Trees, Random Forests, and XGBoost, combined through voting and 5-fold cross-validation, are squared and used as edge weights, and the four community-detection algorithms are then applied to this weighted network. Three node-pair features carry the micro-level information: degree heterogeneity for sparse networks, clustering-coefficient heterogeneity, and common-neighbor count for dense networks. The squaring step converts a probabilistic same-community score into a symmetric similarity weight, which is the object that changes the input graph seen by the detector. In the statistical-physics mode, a preliminary community partition supplies the training labels; in the ground-truth mode, real labels replace that preliminary partition.

What would settle it

Hold out a set of ground-truth communities entirely from training, train the classifier only on the remaining labels, then evaluate NMI and ARI on the held-out communities; if the gain over the original algorithms disappears or shrinks to the statistical-physics mode's gain, the ground-truth mode's advantage comes from label reuse rather than from node-pair similarity.

Watch

Extended reading notes

Core claim

The central claim is that micro-level node-pair similarity is a usable signal, not a side detail: a low-complexity machine-learning step that maps node-pair features to same-community likelihoods, then re-weights the network with the squared likelihoods, makes standard community detectors find partitions that score higher on modularity, NMI, and ARI. When ground-truth communities are known, training the classifier on them and then measuring accuracy against the same labels yields the most accurate recovery in the paper's experiments. When labels are unavailable, using one of the four baseline detectors to produce the training labels still improves modularity and, on several networks, improves NMI and ARI as well. The paper interprets the improvement as a bridge between local node interactions and global network organization, and supports the interpretation with a Pearson-correlation analysis linking the similarity gap to the evaluation metrics.

Load-bearing premise

The ground-truth mode assumes it is legitimate to train the node-pair classifier on the same known community labels that are later used to score the detected communities; if that reuse is information leakage, the reported accuracy gains are partly self-fulfilling.

Editorial extensions

If this is right

  • Any of the four tested detectors—Louvain, Leiden, Infomap, or Fast-Greedy—can be improved by replacing the original graph with the weighted similarity network, so the reported gain is not tied to a single algorithm.
  • When ground-truth communities are available, the supervised mode gives the largest NMI and ARI improvements, with relative gains exceeding 125 percent on some networks.
  • The strong correlation between the internal-versus-external similarity gap and detection metrics implies that node-pair similarity is a meaningful predictor of community membership, not a randomly chosen feature.
  • The framework is low-complexity: it uses three simple topological features and standard ensemble classifiers, so the preprocessing cost stays modest on networks with hundreds of thousands of nodes.

Reading between the lines

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

  • Because the re-weighting step only changes the input graph, the same wrapper could be applied to spectral, stochastic-block-model, or deep-learning community detectors; the paper demonstrates it only with the four classic algorithms.
  • The decision to square each predicted likelihood before treating it as an edge weight is unexamined; testing other monotone transformations (identity, square root, log) would reveal how much of the gain comes from that specific calibration.
  • The correlation evidence points to a direct optimization target—the internal-versus-external similarity gap—that could be trained as a loss function instead of a classification likelihood; the paper does not explore this.
  • The framework could be extended to attributed or temporal networks by adding node attributes or time-respecting similarity features, since the micro-feature set in the paper is purely topological.
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.

Referee Report

5 major / 7 minor

Summary. The paper proposes a community-detection framework that uses machine-learning classifiers on node-pair similarity features (degree heterogeneity, clustering-coefficient heterogeneity, and common neighbors) to assign weights to edges, after which standard algorithms (Louvain, Leiden, Infomap, Fast-Greedy) are applied to the resulting weighted similarity network. Two training regimes are considered: a “statistical-physics” mode in which an initial detection supplies pseudo-labels, and a “ground-truth” mode in which the true community labels are used as training labels. Experiments on one synthetic and five real networks report weighted modularity, NMI, and ARI, along with a correlation analysis between node-pair similarity and these metrics. The abstract claims the framework consistently outperforms conventional methods and is most accurate when ground-truth labels are available.

Significance. The topic is timely and the pipeline is simple enough to be useful if validated; the authors test on several large SNAP datasets and compare against four standard detectors. However, the evidence offered for the central claim is not sound: the ground-truth evaluation is circular because the classifier is trained on the same labels used for scoring, the reported numbers are selected as best-of-16 without correction, no uncertainty estimates are given, and the correlation results are not statistically significant at conventional levels. The paper therefore does not currently substantiate its headline claims. The idea of using machine-learned edge weights informed by mesoscopic structure remains worth exploring, and the modularity improvements in the statistical-physics mode might survive a more careful evaluation, but the present manuscript does not demonstrate that.

major comments (5)
  1. [Section 3, Fig. 3(b), and Section 4.3, Tables 3-4] The “ground-truth” mode trains the node-pair classifier on the exact community labels that are later used to compute NMI and ARI. The five-fold cross-validation shown in Fig. 3(f) is applied to node pairs, not to networks, so every fold shares the same network and the same target partition. The model therefore sees the target labels before the final partition is scored. This makes the reported improvements, such as DBLP VC-FG NMI 0.640 versus Infomap 0.410 and Amazon RF-Infomap ARI 0.342 versus Leiden 0.143, a measure of fit to the training labels rather than independent detection accuracy. The comparison against Louvain, Leiden, Infomap, and Fast-Greedy is thus not a fair comparison among detection methods, and the Abstract's claim that the approach is “most accurate” when ground-truth labels are available is not supported.
  2. [Tables 2-4 captions] Each caption states that the best result among 16 configurations is reported. Selecting the best of 16 outcomes for each dataset and metric, without correction for multiple comparisons and without reporting the variance or the full set of results, inflates the apparent advantage of the proposal. This applies to both the statistical-physics and ground-truth sections and makes the quantitative superiority claims unverifiable. Please report averages and standard deviations over configurations, or pre-specify the configuration choice on a validation set.
  3. [Table 5] The Abstract claims “a strong and statistically significant correlation,” but Table 5 reports p-values of 0.084 and 0.052 for the ground-truth approach and 0.159 and 0.226 for the statistical-physics approach. None of these reaches the 0.05 level, and with only six networks the test has very low power. Moreover, the correlation is computed from the best detection results, which were themselves selected from the same 16 configurations, so the analysis is post-selection. The significance claim must be removed or replaced with a properly powered test.
  4. [Section 4.2 and Fig. 3(b)] In the statistical-physics mode, the initial detector's output is used to label the training node pairs, and the same detectors are then re-run on the reweighted network. If the learned weighting mostly reinforces the teacher's partition, high Qw or NMI relative to the unweighted baseline is an expected self-confirmation rather than evidence of improved detection. A control experiment, such as weighting edges by the teacher's own within-community counts or by random weights matched to the same marginal distribution, is needed to show that the gains are not an artifact of this closed loop.
  5. [Throughout Section 4] No error bars or repeated runs are reported anywhere; for example, the Tree ARI improvement from 0.020 to 0.060 in Table 4 rests on single-point estimates. Since several of the underlying detectors are stochastic and the network-construction procedure in Section 4.1.2 involves random tie-breaking, the reported single numbers are insufficient to establish consistent improvements.
minor comments (7)
  1. [Section 3, Eqs. (3)-(4)] The symbol R_{ij} is used for both degree heterogeneity and clustering-coefficient heterogeneity; please use distinct symbols to avoid confusion.
  2. [Section 4.1.1] The parameter description is confusing: m=1 and n=100 are described as governing links, while Table 1 lists the Tree dataset with 100 nodes and 200 links. Please clarify the generation process and the meaning of each parameter.
  3. [Fig. 2] The label “P-value = 0.00” should be reported as “P < 0.001”, since a p-value cannot equal exactly zero.
  4. [Section 2.3] There is a typo in “communityi” in the sentence following Eq. (2); it should read “community i”.
  5. [Section 5] The concluding paragraph on a “teacher-student-teacher” cycle is speculative and is not tested by any experiment in the paper; it should be removed or supported with evidence.
  6. [Tables 2-4] The captions say “best detection performance among 16 results” but do not define the 16 configurations; please state exactly which classifiers, detectors, feature sets, and voting schemes are included in the search.
  7. [Section 7] The data-availability statement says that code will be released only after acceptance; for a computational paper, providing the code at submission time would materially aid reproducibility.

Circularity Check

3 steps flagged · score 8.0 of 10

Ground-truth mode trains on the exact labels it is scored against; the statistical-physics mode bootstraps from its own first-pass output.

  1. fitted input called prediction [Section 3, Fig. 3 caption; Section 4.3, Tables 3-4]
    "Note that when ground-truth community information is available, step (b) can be omitted. We refer to this approach as the ground-truth method. (c) We extract mesoscopic structural information from detected communities. Specifically, we sample first-order and second-order node pairs, classifying intra-community pairs as one category and inter-community pairs as another. These samples form the training and testing datasets for machine learning."

    In the ground-truth mode, the binary labels used to train the similarity classifier are exactly the ground-truth communities against which NMI and ARI are computed in Section 4.3. The learned edge weights therefore encode the target partition before the final detection is run; Tables 3-4 then report that the same method achieves higher NMI and ARI as if this were independent accuracy. Five-fold cross-validation over node pairs does not remove the leakage, because all folds share the same network and the same target partition. The comparison is thus between detectors that never see the labels and one whose training labels are the scoring labels, making the reported NMI/ARI advantage a measure of fit to the training labels rather than a prediction.

  2. fitted input called prediction [Section 3 Methodology, Fig. 3(b)-(h); Section 4.2, Table 2]
    "In the absence of ground-truth communities, we implement the community detection framework using a statistical-physics approach. (b) We apply four commonly used community detection algorithms (section 2.2) to identify communities within the network, treating these detected communities as the ground truth. (c) We extract mesoscopic structural information from detected communities."

    The teacher partition is produced by the same four algorithms that are later re-run on the similarity-weighted network. The machine-learned weights are trained to reproduce this first-pass partition, and the final detection then re-optimizes on a network whose weights encode that partition. The modularity improvements reported in Table 2 therefore largely measure consistency with the bootstrap partition, not an independent confirmation that node-pair similarity improves detection. This is a closed teacher-student loop rather than an out-of-sample prediction.

1 more flagged steps
  1. other [Section 4.4, Table 5 and caption]
    "To explain the performance improvement of our framework, we analyze the correlation between node-pair similarity and evaluation metrics (NMI and ARI). Specifically, we quantify this correlation using data from step (g) of our framework (Fig. 3(g)), where node-pair similarity and evaluation metrics are derived from the best detection results."

    Sin-Sout is computed on the weighted similarity network whose edge weights were fitted to the same community memberships that define NMI and ARI. Under the ground-truth rule, both the similarity contrast and the accuracy metrics are generated from the same target labels, so the reported r=0.827/0.876 is a self-consistency check of the fitted model, not evidence that similarity independently explains accuracy. The statistical-physics rule has the same issue, with the bootstrap partition serving as the shared reference.

full rationale

The central supervised claim reduces by construction: Section 3 trains the node-pair classifier on the same ground-truth partition that Section 4.3 uses to compute NMI and ARI, so the reported 'most accurate detection results' are agreement with the training labels, not an independent prediction. The unsupervised arm is a bootstrap: a first-pass detection supplies the training communities and the same detectors are re-run on the resulting weighted network, so the modularity gains in Table 2 partly re-amplify the first-pass partition. The correlation analysis in Table 5 is likewise a self-consistency check on the fitted weights. Self-citations to prior feature papers are not the cause; the fitted-input/evaluation loop is internal to the paper's own protocol. Best-of-16 reporting further inflates differences but is secondary to the label leakage.

Assumptions & free parameters 1 free parameters · 4 assumptions · 0 invented entities

No new physical entities are introduced; the framework relies on assumptions about feature sufficiency, sample representativeness, and the reliability of the teacher communities. The squaring of predicted likelihoods is an unexplained, hand-chosen transformation that affects all downstream results.

free parameters (1)
  • Similarity squaring exponent = 2
    Section 3 (Constructing a Weighted Similarity Network) states the predicted likelihood is squared to obtain a similarity weight, with no justification and no search over exponents. This hand-chosen nonlinearity affects the weighted network and the resulting partitions.
assumptions (4)
  • domain assumption The three structural features (degree heterogeneity, clustering coefficient heterogeneity, and common neighbors) are sufficient to predict community co-membership.
    Section 3 (d) introduces exactly these features with no ablation or feature selection; the entire framework rests on their informativeness.
  • domain assumption First-order and second-order node pairs sampled from the network are representative of all pairs relevant to community structure.
    Section 3 (c) states sampling 'first-order and second-order node pairs' without specifying sampling ratios or balance; the classifier's generalization to unsampled pairs is assumed.
  • domain assumption In the statistical-physics mode, the initially detected communities are a reliable teacher for training the classifier.
    Section 3 (b) uses one of the four algorithms to generate training labels in the absence of ground truth; if the initial detection is poor, the whole loop is trained on noise.
  • standard math Standard definitions of modularity, NMI, and ARI are accepted without proof.
    Equations (1)-(2) and metric definitions in Section 2.3 are taken as given from the literature.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Machine Learning Informed by Micro and Mesoscopic Statistical Physics Methods for Community Detection." pith.science (2026). https://pith.science/paper/KE5CCSQV

@misc{pith2026250413538,
  author       = {Pith},
  title        = {Pith review of: Machine Learning Informed by Micro and Mesoscopic Statistical Physics Methods for Community Detection},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/KE5CCSQV}},
  note         = {Machine review of arXiv:2504.13538}
}
read the original abstract

Community detection plays a crucial role in understanding the structural organization of complex networks. Previous methods, particularly those from statistical physics, primarily focus on the analysis of mesoscopic network structures and often struggle to integrate fine-grained node similarities. To address this limitation, we propose a low-complexity framework that integrates machine learning to embed micro-level node-pair similarities into mesoscopic community structures. By leveraging ensemble learning models, our approach enhances both structural coherence and detection accuracy. Experimental evaluations on artificial and real-world networks demonstrate that our framework consistently outperforms conventional methods, achieving higher modularity and improved accuracy in NMI and ARI. Notably, when ground-truth labels are available, our approach yields the most accurate detection results, effectively recovering real-world community structures while minimizing misclassifications. To further explain our framework's performance, we analyze the correlation between node-pair similarity and evaluation metrics. The results reveal a strong and statistically significant correlation, underscoring the critical role of node-pair similarity in enhancing detection accuracy. Overall, our findings highlight the synergy between machine learning and statistical physics, demonstrating how machine learning techniques can enhance network analysis and uncover complex structural patterns.

Figures

Figures reproduced from arXiv: 2504.13538 by the authors.

Figure 1
Figure 1. A real-world network illustrating the limitations of existing community detection algorithms. Nodes [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. The node-pair similarity related to network modularity. The network modularity [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. Overview of the proposed detection framework. In the absence of ground-truth communities, we imple [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: Performance improvement of the proposed framework relative to the original method. Here, [PITH_FULL_IMAGE:figures/full_fig_p010_4.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. Triadic Closure-Heterogeneity-Harmony GCN for Link Prediction

    cs.SI 2025-04 conditional novelty 5.0 of 10

    A GCN variant that injects two classic topology heuristics, common neighbors and degree heterogeneity, into the convolution weights achieves modest AUC gains over plain GCN on nine datasets.

Reference graph

Works this paper leans on

50 extracted references · 41 canonical work pages · cited by 1 Pith paper

  1. [1]

    Link communities reveal multiscale complexity in networks

    Yong-Yeol Ahn, James P Bagrow, and Sune Lehmann. Link communities reveal multiscale complexity in networks. nature, 466(7307):761–764, 2010

  2. [2]

    Xgboost: A scalable tree boosting system

    Tianqi Chen and Carlos Guestrin. Xgboost: A scalable tree boosting system. In Proceedings of the 22nd acm sigkdd international conference on knowledge discovery and data mining, pages 785–794, 2016. 11

  3. [3]

    Finding community structure in very large net- works

    Aaron Clauset, Mark EJ Newman, and Cristopher Moore. Finding community structure in very large net- works. Physical Review E—Statistical, Nonlinear, and Soft Matter Physics, 70(6):066111, 2004

  4. [4]

    Community detection, link prediction, and layer interdependence in multilayer networks

    Caterina De Bacco, Eleanor A Power, Daniel B Larremore, and Cristopher Moore. Community detection, link prediction, and layer interdependence in multilayer networks. Physical Review E, 95(4):042317, 2017

  5. [5]

    Community detection in graphs

    Santo Fortunato. Community detection in graphs. Physics reports, 486(3-5):75–174, 2010

  6. [6]

    Community detection in networks: A user guide.Physics reports, 659:1–44, 2016

    Santo Fortunato and Darko Hric. Community detection in networks: A user guide.Physics reports, 659:1–44, 2016

  7. [7]

    20 years of network community detection

    Santo Fortunato and Mark EJ Newman. 20 years of network community detection. Nature Physics , 18(8):848–850, 2022

  8. [8]

    Evaluating overfit and underfit in models of network community structure

    Amir Ghasemian, Homa Hosseinmardi, and Aaron Clauset. Evaluating overfit and underfit in models of network community structure. IEEE Transactions on Knowledge and Data Engineering , 32(9):1722–1735, 2019

Show all 50 references
  1. [9]

    Stacking models for nearly optimal link prediction in complex networks

    Amir Ghasemian, Homa Hosseinmardi, Aram Galstyan, Edoardo M Airoldi, and Aaron Clauset. Stacking models for nearly optimal link prediction in complex networks. Proceedings of the National Academy of Sciences, 117(38):23393–23400, 2020

  2. [10]

    Detectability thresholds and optimal algorithms for community structure in dynamic networks

    Amir Ghasemian, Pan Zhang, Aaron Clauset, Cristopher Moore, and Leto Peel. Detectability thresholds and optimal algorithms for community structure in dynamic networks. Physical Review X, 6(3):031005, 2016

  3. [11]

    Modularity and stability in ecological communities

    Jacopo Grilli, Tim Rogers, and Stefano Allesina. Modularity and stability in ecological communities. Nature communications, 7(1):12031, 2016

  4. [12]

    One model to rule them all in network science? Proceedings of the National Academy of Sciences, 117(41):25195–25197, 2020

    Roger Guimerà. One model to rule them all in network science? Proceedings of the National Academy of Sciences, 117(41):25195–25197, 2020

  5. [13]

    Community detection in networks without observing edges

    Till Hoffmann, Leto Peel, Renaud Lambiotte, and Nick S Jones. Community detection in networks without observing edges. Science advances, 6(4):eaav1478, 2020

  6. [14]

    Temporal networks

    Petter Holme and Jari Saramäki. Temporal networks. Physics reports, 519(3):97–125, 2012

  7. [15]

    A survey of community detection approaches: From statistical modeling to deep learning

    Di Jin, Zhizhi Yu, Pengfei Jiao, Shirui Pan, Dongxiao He, Jia Wu, S Yu Philip, and Weixiong Zhang. A survey of community detection approaches: From statistical modeling to deep learning. IEEE Transactions on Knowledge and Data Engineering, 35(2):1149–1170, 2021

  8. [16]

    Snap datasets: Stanford large network dataset collection

    Leskovec Jure. Snap datasets: Stanford large network dataset collection. Retrieved December 2024 from http://snap. stanford. edu/data, 2014

  9. [17]

    Network community detection via neural embeddings

    Sadamori Kojaku, Filippo Radicchi, Yong-Yeol Ahn, and Santo Fortunato. Network community detection via neural embeddings. Nature Communications, 15(1):9446, 2024

  10. [18]

    Modularity and dynamics on complex networks

    Renaud Lambiotte and Michael T Schaub. Modularity and dynamics on complex networks . Cambridge University Press, 2021

  11. [19]

    The fundamental advantages of temporal networks

    Aming Li, Sean P Cornelius, Y-Y Liu, Long Wang, and A-L Barabási. The fundamental advantages of temporal networks. Science, 358(6366):1042–1046, 2017

  12. [20]

    Dynamic community detection over evolving networks based on the optimized deep graph infomax

    Hao Liu, Langzhou He, Fan Zhang, Zhen Wang, and Chao Gao. Dynamic community detection over evolving networks based on the optimized deep graph infomax. Chaos: An Interdisciplinary Journal of Nonlinear Science, 32(5), 2022

  13. [21]

    Modularity and community structure in networks

    Mark EJ Newman. Modularity and community structure in networks. Proceedings of the national academy of sciences, 103(23):8577–8582, 2006

  14. [22]

    Finding and evaluating community structure in networks

    Mark EJ Newman and Michelle Girvan. Finding and evaluating community structure in networks. Physical review E, 69(2):026113, 2004

  15. [23]

    Network reconstruction and community detection from dynamics

    Tiago P Peixoto. Network reconstruction and community detection from dynamics. Physical review letters, 123(12):128301, 2019. 12

  16. [24]

    Optimizing parameter search for community detection in time-evolving networks of complex systems

    Lima Dias Pinto, Javier Omar Garcia, Kanika Bansal, et al. Optimizing parameter search for community detection in time-evolving networks of complex systems. Chaos: An Interdisciplinary Journal of Nonlinear Science, 34(2), 2024

  17. [25]

    Information granulation-based community detection for social networks

    Ebin Deni Raj, Gunasekaran Manogaran, Gautam Srivastava, and Yulei Wu. Information granulation-based community detection for social networks. IEEE Transactions on Computational Social Systems , 8(1):122– 133, 2020

  18. [26]

    The maximum capability of a topological feature in link prediction

    Yijun Ran, Xiao-Ke Xu, and Tao Jia. The maximum capability of a topological feature in link prediction. PNAS nexus, 3(3):pgae113, 2024

  19. [27]

    Comparing community structure to characteristics in online collegiate social networks

    Veronica Red, Eric D Kelsic, Peter J Mucha, and Mason A Porter. Comparing community structure to characteristics in online collegiate social networks. SIAM review, 53(3):526–543, 2011

  20. [28]

    Maps of random walks on complex networks reveal community structure

    Martin Rosvall and Carl T Bergstrom. Maps of random walks on complex networks reveal community structure. Proceedings of the national academy of sciences, 105(4):1118–1123, 2008

  21. [29]

    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

  22. [30]

    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

  23. [31]

    A novel metric for community detection

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

  24. [32]

    Local dominance unveils clusters in networks

    Dingyi Shi, Fan Shang, Bingsheng Chen, Paul Expert, Linyuan Lü, H Eugene Stanley, Renaud Lambiotte, Tim S Evans, and Ruiqi Li. Local dominance unveils clusters in networks. Communications Physics , 7(1):170, 2024

  25. [33]

    A comprehensive survey on community detection with deep learning

    Xing Su, Shan Xue, Fanzhen Liu, Jia Wu, Jian Yang, Chuan Zhou, Wenbin Hu, Cecile Paris, Surya Nepal, Di Jin, et al. A comprehensive survey on community detection with deep learning. IEEE Transactions on Neural Networks and Learning Systems, 2022

  26. [34]

    Machine learning approach to detect dynamical states from recurrence measures.Chaos: An Interdisciplinary Journal of Nonlinear Science, 34(4), 2024

    Dheeraja Thakur, Athul Mohan, G Ambika, and Chandrakala Meena. Machine learning approach to detect dynamical states from recurrence measures.Chaos: An Interdisciplinary Journal of Nonlinear Science, 34(4), 2024

  27. [35]

    From louvain to leiden: guaranteeing well- connected communities

    Vincent A Traag, Ludo Waltman, and Nees Jan Van Eck. From louvain to leiden: guaranteeing well- connected communities. Scientific reports, 9(1):1–12, 2019

  28. [36]

    Multilayer stochastic block models reveal the multilayer structure of complex networks

    Toni Valles-Catala, Francesco A Massucci, Roger Guimera, and Marta Sales-Pardo. Multilayer stochastic block models reveal the multilayer structure of complex networks. Physical Review X, 6(1):011036, 2016

  29. [37]

    Scalable spectral clustering for overlapping community detection in large-scale networks

    Hadrien Van Lierde, Tommy WS Chow, and Guanrong Chen. Scalable spectral clustering for overlapping community detection in large-scale networks. IEEE Transactions on Knowledge and Data Engineering , 32(4):754–767, 2019

  30. [38]

    Non-negative matrix factorization for overlapping community detection in directed weighted networks with sparse constraints

    Wenxuan Wang, Jun Meng, Huijia Li, and Jingfang Fan. Non-negative matrix factorization for overlapping community detection in directed weighted networks with sparse constraints. Chaos: An Interdisciplinary Journal of Nonlinear Science, 33(5), 2023

  31. [39]

    Constrained fuzzy community detection by a new modularity optimization framework

    Jing Xiao, Yi-Fan Guo, Yu-Qing He, and Xiao-Ke Xu. Constrained fuzzy community detection by a new modularity optimization framework. IEEE Transactions on Network Science and Engineering, 2024

  32. [40]

    Higher-order fuzzy membership in motif modularity optimization

    Jing Xiao, Ya-Wei Wei, Jing Cao, and Xiao-Ke Xu. Higher-order fuzzy membership in motif modularity optimization. IEEE Transactions on Fuzzy Systems, 2024

  33. [41]

    Overlapping community detection in networks: The state-of-the-art and comparative study

    Jierui Xie, Stephen Kelley, and Boleslaw K Szymanski. Overlapping community detection in networks: The state-of-the-art and comparative study. Acm computing surveys (csur), 45(4):1–35, 2013

  34. [42]

    Defining and evaluating network communities based on ground-truth

    Jaewon Yang and Jure Leskovec. Defining and evaluating network communities based on ground-truth. In Proceedings of the ACM SIGKDD workshop on mining data semantics, pages 1–8, 2012. 13

  35. [43]

    Local higher-order graph clustering

    Hao Yin, Austin R Benson, Jure Leskovec, and David F Gleich. Local higher-order graph clustering. In Proceedings of the 23rd ACM SIGKDD international conference on knowledge discovery and data mining , pages 555–564, 2017

  36. [44]

    An interpretable station delay prediction model based on graph community neural network and time-series fuzzy decision tree

    Dalin Zhang, Yi Xu, Yunjuan Peng, Chenyue Du, Nan Wang, Mincong Tang, Lingyun Lu, and Jiqiang Liu. An interpretable station delay prediction model based on graph community neural network and time-series fuzzy decision tree. IEEE Transactions on Fuzzy Systems, 31(2):421–433, 2022

  37. [45]

    Evaluating accuracy of community detection using the relative normalized mutual information

    Pan Zhang. Evaluating accuracy of community detection using the relative normalized mutual information. Journal of Statistical Mechanics: Theory and Experiment, 2015(11):P11006, 2015

  38. [46]

    Scalable detection of statistically significant communities and hierarchies, using message passing for modularity

    Pan Zhang and Cristopher Moore. Scalable detection of statistically significant communities and hierarchies, using message passing for modularity. Proceedings of the National Academy of Sciences , 111(51):18144– 18149, 2014

  39. [47]

    Detecting net- work communities based on central node selection and expansion

    Zhili Zhao, Nana Zhang, Jiquan Xie, Ahui Hu, Xupeng Liu, Ruiyi Yan, Li Wan, and Yue Sun. Detecting net- work communities based on central node selection and expansion. Chaos, Solitons & Fractals, 188:115482, 2024

  40. [48]

    Hierarchi- cal organization unveiled by functional connectivity in complex brain networks

    Changsong Zhou, Lucia Zemanová, Gorka Zamora, Claus C Hilgetag, and Jürgen Kurths. Hierarchi- cal organization unveiled by functional connectivity in complex brain networks. Physical review letters , 97(23):238103, 2006

  41. [49]

    Robustecd: Enhancement of network structure for robust community detection

    Jiajun Zhou, Zhi Chen, Min Du, Lihong Chen, Shanqing Yu, Guanrong Chen, and Qi Xuan. Robustecd: Enhancement of network structure for robust community detection. IEEE Transactions on Knowledge and Data Engineering, 35(1):842–856, 2021

  42. [50]

    Unsupervised community detection in attributed networks based on mutual information maximization

    Junyou Zhu, Xianghua Li, Chao Gao, Zhen Wang, and Jurgen Kurths. Unsupervised community detection in attributed networks based on mutual information maximization. New Journal of Physics, 23(11):113016, 2021. 14

Pith tools

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