Pith. sign in

REVIEW 2 major objections 5 minor 24 references

A repository for discovery and reuse of higher-order network datasets

T0 review · 2 major / 5 minor · reviewed 2026-08-05 · deepseek-v4-flash

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

Pith's one-line read This paper introduces AHORN, a curated repository that standardizes 83 higher-order network datasets and documents each one's provenance.

desk verdict A useful, clearly written infrastructure paper that needs its live repository snapshot pinned down before acceptance. read the letter →

arxiv 2608.03491 v1 pith:4H2KDRID submitted 2026-08-04 cs.DL cs.SIphysics.data-anphysics.soc-ph

Florian Frantzen , Michael T. Schaub This is my paper · ORCID
classification cs.DLcs.SIphysics.data-anphysics.soc-ph
keywords higher-ordernetworkshypergraphssimplicialcomplexesdatarepositorydatasetprovenancecurationFAIRbenchmarkdatasets
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

Higher-order network data—hypergraphs, simplicial complexes, and related formalisms—are scattered across papers and software-specific collections, making them hard to discover and compare. This paper presents AHORN, a curated repository that converts publicly released source datasets into a standardized line-based format, records provenance and citation guidance, and publishes each versioned release through a browsable catalog and programmatic tooling. The repository currently reports 83 datasets with 94 revisions, spanning social, collaboration, biomedical, and geometric benchmark domains. If the platform and its archived files match the description, AHORN would give the field a common, citable source of higher-order benchmark data that supports reproducible reuse across analysis libraries.

What carries the argument

The central mechanism is the AHORN curation pipeline: source-specific converter scripts normalize each upstream dataset into a canonical, line-based exchange format (version 0.3), with the first line carrying dataset-level JSON metadata and subsequent lines defining nodes or interactions with optional attribute objects. Each submission is validated by the companion tool ahorn-loader before acceptance, then published as a stable, revision-addressable record and linked to provenance, citation guidance, and source code. The combination of committed converters, format validation, versioned archival records, and a machine-readable catalog index is what makes discovery and reproducible reuse possi

What would settle it

Query the machine-readable catalog index described in the paper and count the distinct dataset identifiers and revision entries, then resolve a sample of the archived version records and compare their content with the listed metadata. If the counts differ from 83 datasets and 94 revisions, or a sampled record resolves to different data, the inventory claim is wrong.

Watch

Extended reading notes

Core claim

The paper establishes AHORN as a stand-alone repository layer for higher-order network datasets. Its central claim is that AHORN transforms publicly released source datasets into standardized higher-order network artifacts, documents their provenance, and provides access through both a web catalog and companion programmatic tooling. The curation workflow requires an inspectable upstream source, a committed converter script, and a datasheet-style entry recording provenance, licensing, citation, and limitations; accepted files are released as versioned records so a given revision can be cited and retrieved stably. At the time of writing the catalog contains 83 datasets and 94 revisions, with 7

Load-bearing premise

The claim rests on the live web catalog and its archived dataset files actually containing the reported 83 datasets and 94 revisions with the stated statistics, which cannot be verified from the manuscript text alone.

Editorial extensions

If this is right

  • Researchers can pin a specific revision of a benchmark dataset, so later catalog updates will not silently change the data used in an experiment.
  • A standardized exchange format lets the same dataset be read by different analysis pipelines, with format validation run before release.
  • Citation guidance on each entry points users to the original source publication and the exact released version, making dataset credit and provenance explicit.
  • The catalog's structural statistics allow users to compare datasets before downloading, supporting informed benchmark selection.
  • Stable, versioned releases extend FAIR-style discoverability and interoperability to higher-order network data, which previously lacked such a common layer.

Reading between the lines

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

  • If the reported inventory is accurate and maintained, AHORN could become a de facto standard corpus for higher-order network benchmarking, much as general graph collections are used in dyadic network research.
  • The published structural profiles hint at reusable design guidance: datasets cluster into repeated-event, unique-interaction, and label-imbalanced regimes, so future work could build dataset-recommendation tools from these statistics.
  • The converter-based workflow is inherently extensible; if the format grows to support directed interactions or trajectory data, the same provenance-and-validation pipeline could cover additional classes of higher-order systems.
  • A concrete test of the repository's usefulness would be whether independent groups can reproduce published analyses using only revision-pinned downloads and the documented converters.
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 paper introduces AHORN, a curated repository for higher-order network datasets. It describes the curation workflow (converters, datasheet-style metadata, maintainer review, Zenodo archival), the canonical plain-text exchange format with HIF exports, the web/API and ahorn-loader access routes, validation tooling, and a catalog snapshot of 83 datasets with 94 revisions. The paper reports structural and metadata statistics of the catalog and positions AHORN relative to related repositories such as HypergraphRepository, XGI-DATA, hypergraphx-data, and graph repositories. The contribution is infrastructural rather than methodological; the central claim is that AHORN is a functioning, citable, versioned repository that makes higher-order network datasets easier to discover, compare, and reuse.

Significance. If the repository and catalog are live and as described, this paper is a useful contribution to the higher-order network community. The open-source workflow, machine-readable index, revision-pinned Zenodo records, companion loader, and validation step are concrete and well motivated. The paper is also honest about limitations, including uneven metadata coverage, the parent/child dataset overlap in the catalog counts, and the lack of standardized directed or trajectory data. No mathematical derivation is involved, so soundness rests on whether the described artifact exists and is accessible. The main weakness is that the core empirical claims—83 datasets, 94 revisions, and all derived statistics—are self-reported and tied to mutable live URLs; the paper should make the snapshot independently checkable.

major comments (2)
  1. [§4, §6] The central empirical claims—83 datasets, 94 revisions, and the catalog statistics shown in Figures 2–4—are not verifiable from the manuscript text alone. The URLs in §6 point to mutable live services, and the sentence in §4, 'The complete dataset catalog is provided as supplementary material,' references a supplement that is not part of the visible text. Please include a static inventory table or a versioned, checksummed snapshot of /api/datasets.json in the paper or supplement, listing each dataset's identifier, revision, Zenodo DOI, and key counts. Without such a snapshot, the existence and coverage of the repository cannot be independently checked.
  2. [§4, §5] The headline count '83 datasets' counts parent datasets and derived child datasets (e.g., the combined cooking dataset and cuisine-specific children) as separate entries. The authors acknowledge this in §5, but the abstract and introduction use '83 datasets' without that qualification, so readers may overinterpret the number as the count of independent source datasets. Please report both the inventory-entry count and the number of distinct upstream source datasets, or explicitly phrase the headline as 'published inventory entries.' This clarification is load-bearing for the paper's main coverage claim.
minor comments (5)
  1. [§2.2] The canonical line format is described verbally but no example is shown. A short example of the JSON metadata line, a node line, and an interaction line would make the format much easier to understand without visiting the external URL.
  2. [Table 1] The header row is difficult to parse, particularly the grouping of 'Web API' and 'Dataset Stats.' Consider restructuring the table or adding a clear legend for the check marks and tildes.
  3. [§1.1] The statement that HypergraphRepository is 'currently inaccessible' should include the access date, as the availability of a website can change quickly. The access date is mentioned in the Table 1 footnote, but the main text would benefit from the same precision.
  4. [Figure 4] The caption says 'using Euclidean distance' but the text refers to 'label imbalance.' Please state the formula explicitly or cite the specific equation from reference [23] so the reader does not need to infer the measure.
  5. [§5] The limitation that the format does not yet standardize directed interactions or trajectory data is useful. Consider adding a sentence on whether these are planned for future format versions, as this would help readers assess suitability for their use cases.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the paper is an infrastructural description and its inventory statistics are self-reports, not predictions derived from fitted inputs or self-citation chains.

full rationale

This paper does not contain a mathematical derivation, fitted model, or predictive claim whose output could reduce to its input. The central claim is that AHORN is a functioning curated repository with 83 datasets and 94 revisions (Section 4), supported by descriptions of its architecture, workflow, and access routes. Those inventory counts and the architectural statements describe the artifact itself; they are self-referential in the sense that reporting on a system's own contents is not independent evidence of the system's utility, but they are not circular in the logical sense required by the rubric: there is no equation or construction that makes the claimed result true by definition, and no parameter is fitted to data and then renamed as a prediction. The paper's self-citations (references [3] and [5], which include the second author) appear only in the introduction as illustrative examples of coauthorship and social-contact datasets, and are not load-bearing for any core claim. Claims about live availability (URLs in Sections 2.2 and 6) are externally checkable and falsifiable by visiting the repository or the Zenodo community; a mismatch would weaken the paper's factual support but would not constitute circularity. The absence of a derivation chain means there is nothing to walk backward. The correct finding is therefore no significant circularity.

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

No free parameters; this is an infrastructure paper without mathematical modeling. The main axioms are domain assumptions about licensing, the sufficiency of the exchange format, and the correctness of the catalog statistics.

assumptions (3)
  • domain assumption Upstream datasets are publicly released and their redistribution is compatible with their licenses.
    Section 2.1: candidate datasets must have an identifiable upstream source, sufficient metadata, and redistribution compatible with the upstream license. If this fails for some entries, the repository cannot legally host them.
  • domain assumption The AHORN exchange format v0.3 is a sufficient and correct canonical representation for the curated higher-order datasets.
    Section 2.2: the format is line-based with JSON metadata; multi-network files use JSON marker lines. The utility of the repository depends on this format preserving the necessary semantics.
  • domain assumption The catalog statistics in Section 4 are computed correctly from the repository's inventory.
    The reported counts and metrics (83 datasets, 94 revisions, median shares) are presented without error bars or independent verification; they underlie the coverage claims.

how reviews work

0 comments
Cite this review

Pith. "Pith review of A repository for discovery and reuse of higher-order network datasets." pith.science (2026). https://pith.science/paper/4H2KDRID

@misc{pith2026260803491,
  author       = {Pith},
  title        = {Pith review of: A repository for discovery and reuse of higher-order network datasets},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/4H2KDRID}},
  note         = {Machine review of arXiv:2608.03491}
}
read the original abstract

Higher-order network datasets are dispersed across publications, institutional archives, and software-specific collections, making them difficult to discover, compare, and reuse. We introduce the Aachen Higher-Order Repository of Networks (AHORN), a curated repository of standardized higher-order network datasets derived from publicly released sources. Each dataset entry links a converted dataset to its source, metadata, citation guidance, conversion code, and version history. The repository supports browsable and machine-readable discovery, revision-specific downloads, format validation, and exports for interoperable reuse. We describe the repository architecture, curation workflow, access tools, and the coverage and limitations of the catalog snapshot analyzed in this article.

Figures

Figures reproduced from arXiv: 2608.03491 by the authors.

Figure 1
Figure 1. Overview of the AHORN platform. Contributors transform source datasets into a common exchange format and upload versioned files to Zenodo; maintainers review submissions and publish catalog entries that support discovery, citation, programmatic access through ahorn-loader, and downstream reuse. 1.1 Related Work While substantial repositories for higher-order network data remain scarce, several notable initiatives ha… view at source ↗
Figure 2
Figure 2. Catalog-level distributions across datasets, stratified by network type. The left panel depicts the distribution across application domains, whereas the right panel summarizes metadata availability. for simplicial complexes, 30.0 for hypergraphs, and 81.0 for datasets with multiple types; the single combinatorial complex reaches 315. Moreover, metadata coverage is high for the core discovery fields [PITH_FULL_IMAGE… view at source ↗
Figure 3
Figure 3. Dataset scale and interaction-size distributions across the repository. The Dataset scale panel presents individual datasets, stratified by their number of nodes and interactions, with coloring denoting formalism groups and marker shapes differentiating between temporal and static datasets. The Interaction-size distribution panel illustrates the median per-dataset proportion of interactions as a function of interact… view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: Label imbalance degrees (using Euclidean distance) for all datasets with node or edge labels. 6 Data and Code Availability The AHORN web platform is accessible at https://ahorn.rwth-aachen.de/, with machine-readable discovery metadata available at https://ahorn.rwth-aa…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

24 extracted references · 23 canonical work pages

  1. [1]

    Battiston, F.et al.Networks beyond pairwise interaction: Structure and dynamics.Physics Reports874, 1–92 (2020)

    work page 2020

  2. [2]

    Battiston, F.et al.The physics of higher-order interactions in complex systems.Nature Physics 17, 1093–1098 (2021)

    work page 2021

  3. [3]

    R., Abebe, R., Schaub, M

    Benson, A. R., Abebe, R., Schaub, M. T., Jadbabaie, A. & Kleinberg, J. Simplicial closure and higher-order link prediction.Proceedings of the National Academy of Sciences of the United States of America115, E11221–E11230 (2018)

    work page 2018

  4. [4]

    A., Saucan, E

    Murgas, K. A., Saucan, E. & Sandhu, R. Hypergraph geometry reflects higher-order dynamics in protein interaction networks.Scientific Reports12, 20879 (2022)

    work page 2022

  5. [5]

    & Schaub, M

    Neuhäuser, L., Lambiotte, R. & Schaub, M. T. Consensus Dynamics and Opinion Formation on Hypergraphs. In Battiston, F. & Petri, G. (eds.)Higher-Order Systems, 347–376 (Springer International Publishing, Cham, 2022)

    work page 2022

  6. [6]

    D.et al.The FAIR Guiding Principles for scientific data management and stewardship.Scientific Data3, 160018 (2016)

    Wilkinson, M. D.et al.The FAIR Guiding Principles for scientific data management and stewardship.Scientific Data3, 160018 (2016). 7

    work page 2016

  7. [7]

    & Spagnuolo, C

    Antelmi, A., De Vinco, D. & Spagnuolo, C. HypergraphRepository: A Community-Driven and Interactive Hypernetwork Data Collection. In Dewar, M.et al.(eds.)Modelling and Mining Networks, vol. 14671, 159–173 (Springer Nature Switzerland, Cham, 2024)

    work page 2024

  8. [8]

    W.et al.XGI: A Python package for higher-order interaction networks.Journal of Open Source Software8, 5162 (2023)

    Landry, N. W.et al.XGI: A Python package for higher-order interaction networks.Journal of Open Source Software8, 5162 (2023)

    work page 2023

Show all 24 references
  1. [9]

    F.et al.Hypergraphx: A library for higher-order network analysis.Journal of Complex Networks11, cnad019 (2023)

    Lotito, Q. F.et al.Hypergraphx: A library for higher-order network analysis.Journal of Complex Networks11, cnad019 (2023)

    2023

  2. [10]

    F., Betti, L., Nortier, B., Montresor, A

    Lotito, Q. F., Betti, L., Nortier, B., Montresor, A. & Battiston, F. Hypergraphx-data: A repository for higher-order network data.Journal of Complex Networks14, cnag014 (2026)

    2026

  3. [11]

    Benson, A. R. Austin R. Benson datasets. https://www.cs.cornell.edu/~arb/data/

  4. [12]

    KONECT: The Koblenz network collection

    Kunegis, J. KONECT: The Koblenz network collection. InProceedings of the 22nd International Conference on World Wide Web, 1343–1350 (ACM, Rio de Janeiro Brazil, 2013)

    2013

  5. [13]

    & Krevl, A

    Leskovec, J. & Krevl, A. SNAP Datasets: Stanford large network dataset collection. http://snap.stanford.edu/data (2014)

    2014

  6. [14]

    & Sainz, M

    Clauset, A., Tucker, E. & Sainz, M. The Colorado Index of Complex Networks. https://icon.colorado.edu (2016)

    2016

  7. [15]

    Rossi, R. A. & Ahmed, N. K. An Interactive Data Repository with Visual Analytics.ACM SIGKDD Explorations Newsletter17, 37–41 (2016)

    2016

  8. [16]

    Peixoto, T. P. The Netzschleuder network catalogue and repository (2020)

    2020

  9. [17]

    InThe Thirteenth International Conference on Learning Representations(2025)

    Ballester, R.et al.MANTRA: The manifold triangulations assemblage. InThe Thirteenth International Conference on Learning Representations(2025)

    2025

  10. [18]

    Gebru, T.et al.Datasheets for Datasets.Communications of the ACM64, 86–92 (2021)

    2021

  11. [19]

    Joint Declaration of Data Citation Principles

    Data Citation Synthesis Group. Joint Declaration of Data Citation Principles. Tech. Rep., Force11 (2014)

    2014

  12. [20]

    Coll, M.et al.HIF: The hypergraph interchange format for higher-order networks (2025)

    2025

  13. [21]

    Praggastis, B.et al.HyperNetX: A Python package for modeling complex network data as hypergraphs.Journal of Open Source Software9, 6016 (2024)

    2024

  14. [22]

    Spagnuolo, C.et al.Analyzing, Exploring, and Visualizing Complex Networks via Hypergraphs using SimpleHypergraphs.jl.Internet Mathematics(2020)

    2020

  15. [23]

    & Lozano, J

    Ortigosa-Hernández, J., Inza, I. & Lozano, J. A. Measuring the class-imbalance extent of multi-class problems.Pattern Recognition Letters98, 32–38 (2017)

    2017

  16. [24]

    Journal of Machine Learning Research25, 1–8 (2024)

    Hajij, M.et al.TopoX: A Suite of Python Packages for Machine Learning on Topological Domains. Journal of Machine Learning Research25, 1–8 (2024). Funding The authors acknowledge funding by the European Union (ERC, HIGH-HOPeS, 101039827). Views and opinions expressed are howeve...

    2024

Pith tools

Reviewed August 5, 2026 · model on record in the stance chip above.