Pith. sign in

REVIEW 2 major objections 5 minor 57 references

Hypermodularity and community detection in hypergraphs

T0 review · 2 major / 5 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read Hypergraph community detection reduces to one singular-vector cut of a modularity tensor.

desk verdict Useful spectral heuristic for hypergraph bisection, but the paper's exact-optimality claim for the leading singular vector does not hold up. read the letter →

arxiv 2412.06935 v2 pith:KN73P6BB submitted 2024-12-09 physics.soc-ph cond-mat.dis-nncs.SI

Charo I. del Genio This is my paper · ORCID
classification physics.soc-phcond-mat.dis-nncs.SI MSC 05C6505C8215A69 PACS 89.75.Fb89.75.Hc
keywords hypermodularityhypergraphcommunitydetectionspectralbisectionhigher-ordersingularvaluedecompositionmodularitytensorspinvariablesk-uniformhypergraphsschoolcontactnetworks
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 seeks to put community detection in hypergraphs on the same spectral footing as classical network modularity. The author defines a hypermodularity function and derives a closed combinatorial expression that rewrites it in vector form, $Q = (1/(2^{k-1} k! m)) s^T E \sigma^{(k)}$, so that the search for the best two-community partition becomes a singular-value problem on a flattened modularity tensor. The central claim is that the assignment read off the first left singular vector of that matrix is the true maximizer of hypermodularity, not merely an approximation. If that holds, higher-order community detection inherits the speed of spectral methods while preserving every hyperedge, and the school-contact validations suggest the detected modules track real social and cognitive groupings.

What carries the argument

The load-bearing object is the modularity tensor $B$, whose entries subtract the expected number of hyperedges under a degree-preserving null model from the observed indicator, together with its flattening $E$: a matrix whose rows are nodes and whose columns index all ordered $(k-1)$-tuples. The machinery also includes the spin-like variables $s_i \in \{+1,-1\}$ and the closed-form vector $\sigma^{(k)}$, whose entries are sums of ordered $r$-choices of the spins with ordinal-index coefficients; this turns the non-quadratic, delta-laden hypermodularity expression into the bilinear form $s^T E \sigma^{(k)}$. The first left singular vector of $E$ then supplies the bisection, and the corrected tensor $B'$ is the device that keeps recursive bisection valid when applied to a sub-community rather than the whole network.

What would settle it

Take a small $k$-uniform hypergraph, enumerate every bipartition by exhaustive search to find the true hypermodularity maximum, then compute the first left singular vector of $E$ and read off the sign cut; if any other bipartition has higher hypermodularity than the sign cut, the paper's central exactness claim is false.

Watch

Extended reading notes

Core claim

The discovery is a reformulation: for a simple $k$-uniform hypergraph, the hypermodularity of a bipartition can be written exactly as $Q = (1/(2^{k-1} k! m)) s^T E \sigma^{(k)}$, where $s$ carries $+1/-1$ community labels, $E$ is any standard flattening of the hypersymmetric modularity tensor $B$ (observed hyperedges minus a degree-product null model), and $\sigma^{(k)}$ is built from ordered $r$-choices of $k-1$ spin variables with ordinal-index coefficients. Because the sum of all elements of $B$ vanishes, the constant terms in the chain product of Kronecker deltas drop out, which is what makes the vector form possible. The paper then argues that the sign pattern of the first left singular vector of $E$ gives the bipartition that maximizes hypermodularity, and it proves that all standard flattenings are identical, so no arbitrary mode choice enters. For further splits, a corrected subtensor $B'$ restores the vanishing-sum condition, and repeated bisection with node-level and community-level refinement steps yields multi-community partitions. The method is demonstrated on random $k$-uniform hypergraphs, where connected random instances stay below hypermodularity about $0.2$, and on primary-school and high-school contact data, where the communities found at each edge size align with age groups, classes, and academic subjects.

Load-bearing premise

The method stands or falls with the claim that the partition obtained by cutting the sign pattern of the first left singular vector of the flattened modularity tensor truly maximizes hypermodularity, even though the objective is a cubic-or-higher function of the assignment vector.

Editorial extensions

If this is right

  • If the exactness claim is right, the bisection step of hypergraph community detection is solvable by a single power iteration on $E E^T$, avoiding the NP-hard general maximization of modularity.
  • Recursive bisection with the corrected subtensor $B'$ gives a complete multi-community algorithm, so a hypergraph can be partitioned at every edge size $k$ without projecting hyperedges onto pairwise links.
  • Connected random $k$-uniform hypergraphs have physiological maximum hypermodularity below about $0.2$; values above that on connected data signal genuine community structure, while high values on disconnected networks are artefacts of fragmentation and should be handled per connected component.
  • On the school contact data, the communities found at each $k$ from $2$ to $5$ have natural interpretations in terms of age groups, sibling ties, and shared academic interests, showing that per-order analysis extracts information that pairwise analysis misses.

Reading between the lines

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

  • The exactness claim implies a testable equivalence: any other search procedure that maximizes the same hypermodularity function should never beat the singular-vector cut on the first bisection. Small hypergraphs where exhaustive enumeration is possible would settle this directly.
  • If the singular-vector cut is exact, the same closed combinatorial 'vector form' trick may extend to other higher-order objectives, such as hypergraph modularity density or weighted mixtures of edge sizes, opening a family of spectral higher-order clustering methods.
  • The empirical $0.2$ random-network threshold invites an analytic calculation: as a function of $k$ and the degree sequence, the expected maximum hypermodularity of a random $k$-uniform hypergraph could be derived from the singular-value distribution of the flattened tensor, turning the threshold into a quantitative null model.
  • The paper treats a non-uniform hypergraph as a union of uniform sub-hypergraphs; an implicit extension is a combined objective that weights each edge size, which would let one detect communities visible only through a mixture of interaction orders.
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

2 major / 5 minor

Summary. The manuscript introduces hypermodularity as a quality function for community detection in k-uniform hypergraphs and derives a vector-form expression Q = (1/(2^{k-1} k! m)) s^T E sigma^(k), where E is a flattening of the hypermodularity data tensor and sigma^(k) is a nonlinear function of the partition vector s. The paper claims that the bipartition maximizing hypermodularity is obtained exactly from the first left singular vector of E, and it presents an algorithm combining spectral bisection with Kernighan-Lin, node-level, and community-level refinement steps. The method is tested on synthetic random hypergraphs and on primary-school and high-school contact data for hyperedges of sizes 2 through 5, with qualitative interpretations of the detected communities. Appendices A-C contain the algebraic derivations, and Appendix D gives a small worked example.

Significance. If the exact-optimality claim were true, the paper would supply a remarkable bridge between tensor singular value decomposition and exact modularity maximization for hypergraphs, and would also provide a theoretical justification for HOSVD-based classification. The algebraic rewriting of hypermodularity in vector form is elegant, and the author has made code and data openly available, which is a genuine strength. However, the central exactness claim is not supported by the derivation, as explained below; at present the contribution is best read as a spectral heuristic with interesting real-world illustrations. The paper would be useful in that weaker form, but its significance is substantially reduced until the theoretical claim is either proved or explicitly withdrawn.

major comments (2)
  1. [§II.B (after Eq. 4) and Appendix B (final paragraph)] The statement that the first left singular vector of the flattening E yields the partition that maximizes hypermodularity, called in Appendix B 'indeed the best one, and not just an approximation of it', is not established and is generally false for k >= 3. In Eq. (7)/(B11), Q = (1/(2^{k-1} k! m)) s^T E sigma^(k), but sigma^(k) is a nonlinear function of s of degree k-1 (Eq. 9/B12). Hence Q is a degree-k function of the partition vector, while the first left singular vector of E maximizes the quadratic form ||E^T u||_2 over unit-norm real vectors u. That is a spectral relaxation of the discrete problem, not the original objective. For k=2 the two problems coincide because sigma^(2)=s, but for k>=3 no argument connects the sign pattern of the leading singular vector to the maximizer of the degree-k objective. The worked example in Appendix D illustrates the method but does not fill this gap. I recommend rewriting the relevant passages to describe the spectral step as a relaxation/heuristic, unless a proof is supplied.
  2. [Appendix C, Eq. (12)/(C6)] The repeated-bisection correction defines a modified tensor B' that is claimed to restore the vanishing-sum property needed for the vector form. The paper does not explicitly verify that the total sum of B' is zero; this is load-bearing because the constant term in the product of Kronecker deltas is dropped precisely when the sum vanishes. The property does follow, since the sum of the diagonal corrections over all v equals the total sum of B, which vanishes by construction, but it should be stated and proved explicitly before Eq. (11) is used.
minor comments (5)
  1. [§II.D, Fig. 1] The value 0.2 is presented as a 'physiological expected maximum hypermodularity' of random hypergraphs, but it is estimated from the author's own algorithm on a limited set of sizes (N=10,20,50,100; k=3,4) with no confidence intervals or statistical tests. This threshold is later used to interpret real-network results (e.g., q=0.1723 for a high-school k=5 component), so the claim should be softened or supported by a more systematic null-model analysis.
  2. [General] No quantitative comparison with existing hypergraph community-detection methods (e.g., Refs. 32, 33, 40, 41) is provided. Since the paper claims a new method, benchmark comparisons would considerably strengthen the validation.
  3. [§II.B, Eq. (9)/(B12)] The definition of sigma^(k) is difficult to parse; in particular, the 'inverse lexicographic order' and the ordinal index alpha_1 are not illustrated. A short example for k=3 or k=4 showing the correspondence between the entries of sigma^(k) and the columns of E would greatly improve readability.
  4. [Appendix B, final paragraph] The concluding remark that the result 'provides an explanation of the success of methods based on higher-order SVD in machine learning' is a broad extrapolation that goes beyond the scope of the derivation and should be removed unless it is made precise.
  5. [General] The paper does not discuss the computational complexity of the algorithm, particularly the cost of forming and multiplying by the flattening E of size N x N^{k-1}. A complexity statement would help readers assess scalability.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the hypermodularity objective is independently defined, the vector-form rewrite is an exact algebraic identity, and no prediction is fitted to its own input.

full rationale

The paper's core claim is a spectral method for hypermodularity maximization. The hypermodularity definition in Eq. (2) is taken from previous independent work (Refs. [40,41]) and is acknowledged as mathematically equivalent, not renamed as a new result. The transformation to Eq. (7) is an exact rearrangement: σ^(k) is explicitly defined in Eqs. (8) and (9) in terms of the spin variables, and the constant terms are removed by the stated property that the sum of all elements of B vanishes by construction. This is a derivation, not a definitional round-trip. The assertion in Appendix B that the first left singular vector gives 'the best one, and not just an approximation' is not justified by the algebra, since hypermodularity is a degree-k function of s while the singular vector maximizes a quadratic relaxation; however, this is an unsupported optimality claim, not circularity. No parameter is fitted to a subset of data and then presented as a prediction of a closely related quantity. The 0.2 threshold for random hypergraphs is an empirical baseline measured on independently generated Erdős–Rényi ensembles and is then used only as a rough interpretive reference for real-world values; this is calibration against synthetic benchmarks, not a fitted-input prediction. The self-citations in the paper (Refs. [26,29,47]) support refinement heuristics and a review context, but they are not load-bearing for the central derivation, and no uniqueness theorem is imported from the author's own prior work to force the choice of formalism. The exact-optimality overclaim is a correctness or rigor concern, but it does not reduce the derivation to its inputs.

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

The model introduces no fitted scientific parameters: the hypermodularity coefficients are fixed by the Chung-Lu null model and the degree sequence. The main external assumptions are simplicity of the hypergraph, reduction to uniform hypergraphs, and the unproven spectral optimality premise. No new entities are postulated.

assumptions (6)
  • standard math The sum of all elements of the data tensor B vanishes by construction, allowing constant terms to be dropped.
    Used in Appendix B to derive the vector equation; follows from the null model and the definition of m.
  • domain assumption The hypergraph is simple, meaning no multiple edges, so the adjacency tensor can be written as an indicator.
    Stated at the start of Section II B; excludes multihypergraphs.
  • domain assumption Every hypergraph can be analyzed as a union of independent uniform subhypergraphs, so it suffices to solve the k-uniform case.
    Stated in Section II A; underlies all subsequent equations.
  • ad hoc to paper The first left singular vector of the flattening E yields the partition maximizing hypermodularity.
    Asserted in Section II B and Appendix B without proof; this is the main unsupported premise.
  • domain assumption The expected number of hyperedges in the null model is (k-1)!/(km)^{k-1} times the product of degrees, taken from Refs [40,41].
    This is the Chung-Lu style null model used in Eq. (2); imported from prior work.
  • ad hoc to paper The diagonal correction in Eq. (12) preserves the vanishing-sum property during repeated bisections.
    Introduced in Section II B and derived in Appendix C; the derivation is convoluted and the correction is designed to make the algebra work.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Hypermodularity and community detection in hypergraphs." pith.science (2026). https://pith.science/paper/KN73P6BB

@misc{pith2026241206935,
  author       = {Pith},
  title        = {Pith review of: Hypermodularity and community detection in hypergraphs},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/KN73P6BB}},
  note         = {Machine review of arXiv:2412.06935}
}
read the original abstract

Numerous networked systems feature a structure of nontrivial communities, which often correspond to their functional modules. Such communities have been detected in real-world biological, social and technological systems, as well as in synthetic models thereof. While much effort has been devoted to developing methods for community detection in traditional networks, the study of community structure in networks with higher-order interactions is still not as extensively explored. In this article, we introduce a formalism for the hypermodularity of higher-order networks that allows us to use spectral methods to detect community structures in hypergraphs. We apply this approach to synthetic random networks as well as to real-world data, showing that it produces results that reflect the nature and the dynamics of the interactions modelled, thereby constituting a valuable tool for the extraction of hidden information from complex higher-order data sets.

Figures

Figures reproduced from arXiv: 2412.06935 by the authors.

Figure 1
Figure 1. Maximum hypermodularity in random hypergraphs is never above 0.2. So long as the networks remain connected, the average maximum hypermodularity ⟨Q⟩ of random k-uniform hypergraphs remains below 0.2, as shown here for k = 3 (left-hand top panel) and k = 4 (right-hand top panel). As the connection probability p decreases, the networks start to fragment, and the value of the maximum hypermodularity spuriously increases… view at source ↗
Figure 2
Figure 2. Higher-order communities of primary-school pupils reveal underlying social dynamics. The communities of 2-people and 3-people interactions (top-left and top-right panels, respectively) show that tight-group interactions between primary-school pupils mostly occur within age groups. Communities of looser groups of 4 people (bottom-left) blur this line, with younger siblings being accepted within groups of older childr… view at source ↗
Figure 3
Figure 3. Higher-order communities of high-school students provide insights into closeness of interests. The communities of dyadic interactions between high-school students (top-left) show a strong division between topics (colours) and classes (shades), with the exception of mathematics and physics students (violet), who interact across classes, challenging the stereotypical opinion of more mathematically minded people being … view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: Worked example of the hypermodularity maximization algorithm. (a) The network is a 12-node 3-uniform hypergraph consisting of three cliques, specifically on nodes 1–5, 5–8 and 9–12, joined by edges (3, 4, 7), (4, 5, 7), (6, 8, 11) and (8, 9, 11). All the edges within e…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

57 extracted references · 53 canonical work pages

  1. [1]

    Albert and A.-L

    R. Albert and A.-L. Barabási, Statistical mechanics of complex networks, Rev. Mod. Phys.74, 47 (2002)

    work page 2002

  2. [2]

    M. E. J. Newman, Structure and function of complex networks, SIAM Rev.45, 167 (2003)

    work page 2003

  3. [3]

    Boccaletti, V

    S. Boccaletti, V. Latora, Y. Moreno, M. Chavez and D.- U. Hwang, Complex networks: structure and dynamics, Phys. Rep.424, 175 (2006)

    work page 2006

  4. [4]

    Boccaletti, G

    S. Boccaletti, G. Bianconi, R. Criado, C. I. del Genio, J. Gómez-Gardeñes, M. Romance, I. Sendiña-Nadal, Z. Wang and M. Zanin, The structure and dynamics of mul- tilayer networks, Phys. Rep.544, 1 (2014)

    work page 2014

  5. [5]

    S. L. Pimm, Structure of food webs, Theor. Popul. Biol. 16, 144 (1979)

    work page 1979

  6. [6]

    G. P. Garnett, J. P. Hughes, R. M. Anderson, B. P. Stoner, S. O. Aral, W. L. Whittington, H. H. Hands- field and K. K. Holmes, Sexual mixing patterns of pa- tients attending sexually transmitted diseases clinics, Sex. Transm. Dis.23, 248 (1996)

    work page 1996

  7. [7]

    G. W. Flake, S. Lawrence, C. L. Giles and F. M. Coetzee, Self-organization and identification of web communities, Computer 32, 66 (2002)

    work page 2002

  8. [8]

    K. A. Eriksen, I. Simonsen, S. Maslov and K. Sneppen, Modularity and extreme edges of the internet, Phys. Rev. Lett. 90, 148701 (2003)

    work page 2003

Show all 57 references
  1. [9]

    A.E.Krause, K.A.Frank, D.M.Mason, R.E.Ulanowicz and W. W. Taylor, Compartments revealed in food-web structure, Nature426, 282 (2003)

    2003

  2. [10]

    Lusseau and M

    D. Lusseau and M. E. J. Newman, Identifying the role that animals play in their social networks, Proc. R. Soc. Lond. B Biol.271, S477 (2004)

    2004

  3. [11]

    R.GuimeràandL.A.N.Amaral, Functionalcartography of complex metabolic networks, Nature433 895 (2005)

    2005

  4. [12]

    Palla, I

    G. Palla, I. Derényi, I. Farkas and T. Vicsek, Uncov- ering the overlapping community structure of complex 16 networks in nature and society, Nature435, 814 (2005)

    2005

  5. [13]

    Huss and P

    M. Huss and P. Holme, Currency and commodity metabolites: their identification and relation to the mod- ularity of metabolic networks, IET Syst. Biol. 1, 280 (2007)

    2007

  6. [14]

    486, 75 (2010)

    S.Fortunato, Communitystructureingraphs, Phys.Rep. 486, 75 (2010)

    2010

  7. [15]

    H.Cherifi, G.Palla, B.K.SzymanskiandX.Lu, Oncom- munity structure in complex networks: challenges and opportunities, Appl. Netw. Sci.4, 117 (2019)

    2019

  8. [16]

    M. E. J. Newman, Modularity and community struc- ture in networks, Proc. Natl. Acad. Sci. USA103, 8577 (2006)

    2006

  9. [17]

    Brandes, D

    U. Brandes, D. Delling, M. Gaertler, R. Görke, M. Hoe- fer, Z. Nikoloski and D. Wagner, IEEE T. Knowl. Data En. 20, 172 (2008)

    2008

  10. [18]

    M. Chen, K. Kuzmin and B. K. Szymanski, Community detection via maximization of modularity and its vari- ants, IEEE Trans. Comput. Soc. Syst.1, 46 (2004)

    2004

  11. [19]

    Duch and A

    J. Duch and A. Arenas, Community detection in complex networks using extremal optimization, Phys. Rev. E72, 027104 (2005)

    2005

  12. [20]

    V. D. Blondel, J. L. Guillaume, R. Lambiotte and E. Lefebvre, Fast unfolding of communities in large net- works, J. Stat. Mech. Theory E., P10008 (2008)

    2008

  13. [21]

    Noack and R

    A. Noack and R. Rotta, Multi-level algorithms for mod- ularity clustering, Lect. Notes Comput. Sci.5526, 257 (2009)

    2009

  14. [22]

    Y. Sun, B. Danila, K. Josić and K. E. Bassler, Im- proved community structure detection using a modified fine-tuning strategy, EPL86, 2009

    2009

  15. [23]

    B. H. Good, Y.-A. de Montjoye and A. Clauset, Perfor- mance of modularity maximization in practical contexts, Phys. Rev. E81, 046106 (2010)

    2010

  16. [24]

    Le Martelot and C

    E. Le Martelot and C. Hankin, Multi-scale community detection using stability as optimisation criterion in a greedy algorithm, Proc. Int. Conf. Knowledge Discovery and Information Retrieval (KDIR 2011), 216 (2011)

    2011

  17. [25]

    Sobolevsky, R

    S. Sobolevsky, R. Campari, A. Belyi and C. Ratti, Gen- eral optimization technique for high-quality community detection in complex networks, Phys. Rev. E90, 012811 (2014)

    2014

  18. [26]

    Treviño III, A

    S. Treviño III, A. Nyberg, C. I. del Genio and K. E. Bassler, Fast and accurate determination of modular- ity and its effect size, J. Stat. Mech. Theory E., P02003 (2015)

    2015

  19. [27]

    X. Lu, B. Cross and B. Szymanski, Asymptotic resolu- tion bounds of generalized modularity and multi-scale community detection, Inform. Sciences525, 54 (2020)

    2020

  20. [28]

    Battiston, G

    F. Battiston, G. Cencetti, I. Iacopini, V. Latora, M. Lu- cas, A. Patania, J.-G. Young and G. Petri, Networks beyond pairwise interactions: structure and dynamics, Phys. Rep.874, 1 (2020)

    2020

  21. [29]

    Boccaletti, P

    S. Boccaletti, P. De Lellis, C. I. del Genio, K. Alfaro- Bittner, R. Criado, S. Jalan and M. Romance, The struc- ture and dynamics of networks with higher order inter- actions, Phys. Rep.1018, 1 (2023)

    2023

  22. [30]

    Berge, Graphs and hypergraphs (Elsevier, 1973)

    C. Berge, Graphs and hypergraphs (Elsevier, 1973)

    1973

  23. [31]

    Higher-order systems

    A. Eriksson, T. Carletti, R. Lambiotte, A. Rojas and M. Rosvall, Flow-based community detection in hyper- graphs, in F. Battiston and G. Petri (eds.), “Higher-order systems” (Springer, 2022)

    2022

  24. [32]

    Kumar, S

    T. Kumar, S. Vaidyanathan, H. Ananthapadmanabhan, S. Parthasarathy and B. Ravindran, Hypergraph cluster- ing by iteratively reweighted modularity maximization, Appl. Net. Sci.5, 52 (2020)

    2020

  25. [33]

    Contreras-Aso, R

    G. Contreras-Aso, R. Criado, G. Vera de Salas and J. Yang, Detecting communities in higher-order networks by using their derivative graphs, Chaos Soliton. Fract. 177, 114200 (2023)

    2023

  26. [34]

    J. C. Wright Billings, M. Hu, G. Lerda, A. N. Medvedev, F. Mottes, A. Onicas, A. Santoro and G. Petri, Sim- plex2Vec embeddings for community detection in simpli- cial complexes, arXiv:1906.09068

    1906 arXiv

  27. [35]

    Zhen and J

    Y. Zhen and J. Wang, Community detection in general hypergraph via graph embedding, J. Am. Stat. Ass.118, 1620 (2023)

    2023

  28. [36]

    M. C. Angelini, F. Caltagirone, F. Krzakala and L. Zde- borová, Spectral detection on sparse hypergraphs, Proc. 2015 53rd Ann. Allerton Conf. Commun. Control Comp., 66 (2015)

    2015

  29. [37]

    Rabanser, O

    S. Rabanser, O. Shchur and S. Günnemann, Introduc- tion to tensor decompositions and their applications in machine learning, arXiv:1711.10781

    arXiv

  30. [38]

    Z. T. Ke, F. Shi and D. Xia, Community detection for hypergraph networks via regularized tensor power itera- tion, arXiv:1909.06503

    1909 arXiv

  31. [39]

    Krishnagopal and G

    S. Krishnagopal and G. Bianconi, Spectral detection of simplicial communities via Hodge Laplacians, Phys. Rev. E 104, 064303 (2021)

    2021

  32. [40]

    Kamiński, V

    B. Kamiński, V. Poulin, P. Prałat, P. Szufel and F. Théberge, Clustering via hypergraph modularity, PLoS One 14, e0224307 (2019)

    2019

  33. [41]

    Kamiński, P

    B. Kamiński, P. Misiorek, P. Prałat and F. Théberge, Modularity based community detection in hypergraphs, J. Compl. Netw.12, cnae041 (2024)

    2024

  34. [42]

    Rajwade, A

    A. Rajwade, A. Rangarajan and A. Banerjee, Image de- noising using the higher order singular value decomposi- tion, IEEE T. Pattern Anal.35, 849 (2013)

    2013

  35. [43]

    P.Sankaranarayanan, T.E.Schomay, K.A.AielloandO. Alter, Tensor GSVD of patient- and platform-matched tumor and normal DNA copy-number profiles uncov- ers chromosome arm-wide patterns of tumor-exclusive platform-consistent alterations encoding for cell transfor- mation and pred...

    2015

  36. [44]

    Y.-H. Taguchi, Identification of candidate drugs using tensor-decomposition-based unsupervised feature extrac- tion in integrated analysis of gene expression between diseases and DrugMatrix datasets, Sci. Rep. 7, 13733 (2017)

    2017

  37. [45]

    Taguchi, Tensor decomposition-based unsuper- vised feature extraction applied to matrix products for multi-view data processing, PLoS One 13, e0183933 (2017)

    Y.-H. Taguchi, Tensor decomposition-based unsuper- vised feature extraction applied to matrix products for multi-view data processing, PLoS One 13, e0183933 (2017)

    2017

  38. [46]

    Vilas Boas, W

    B. Vilas Boas, W. Zirwas and M. Haardt, Deep-LaRGE: higher-order SVD and deep learning for model order se- lection in MIMO OFDM systems, Proceedings of the 26th International ITG Workshop on Smart Antennas and 13th Conference on Systems, Communications, and Coding, Braunschwei...

    2023

  39. [47]

    Botta and C

    F. Botta and C. I. del Genio, Finding network communi- ties using modularity density, J. Stat. Mech. Theory E., 123402 (2016)

    2016

  40. [48]

    B. W. Kernighan and S. Lin, An efficient heuristic pro- cedure for partitioning graphs, Bell Syst. Tech. J.49 291 (1970)

    1970

  41. [49]

    Stehlé, N

    J. Stehlé, N. Voirin, A. Barrat, C. Cattuto, L. Isella, J.- 17 F. Pinton, M. Quaggiotto, W. Van den Broeck, C. Régis, B. Lina and P. Vanhems, High-Resolution Measurements of Face-to-Face Contact Patterns in a Primary School, PLoS One6, e23176 (2011)

    2011

  42. [50]

    P. S. Chodrow, N. Veldt and A. R. Benson, Generative hypergraph clustering: From blockmodels to modularity, Sci. Adv.7, eabh1303 (2021)

    2021

  43. [51]

    Mastrandrea, J

    R. Mastrandrea, J. Fournet and A. Barrat, Contact Pat- terns in a High School: A Comparison between Data Collected Using Wearable Sensors, Contact Diaries and Friendship Surveys, PLoS One10, e0136497 (2015)

    2015

  44. [52]

    Analyzing Contemporary Fer- tility

    C. R. Schwartz, C. Doren and A. Li, Trends in Years Spent as Mothers of Young Children: The Role of Com- pleted Fertility, Birth Spacing, and Multiple Partner Fer- tility, in R. Schoen (ed.), “Analyzing Contemporary Fer- tility” (Springer, 2020)

    2020

  45. [53]

    Guimerà, M

    R. Guimerà, M. Sales-Pardo and L. Amaral, Modularity from fluctuations in random graphs and complex net- works, Phys. Rev. E70, 025101(R) (2004)

    2004

  46. [54]

    M. Park, D. W. Feng, S. Digra, T.-A. Vu-Le, L. Anne, G. Chacko and T. Warnow, Improved community detection using stochastic block models, arXiv:2502.00686

    arXiv

  47. [55]

    Fortunato and M

    S. Fortunato and M. Barthélemy, Resolution limit in community detection, Proc. Natl. Acad. Sci. USA104, 36 (2006)

    2006

  48. [56]

    https://codeberg.org/paraw/HyperMod

  49. [57]

    https://charodelgenio.weebly.com/ community-detection-in-higher-order-networks. html

Pith tools

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