{"id":"cf5415cf-9f1c-462e-810e-18e819c3d426","arxiv_id":"2608.03491","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"unknown","formal_verification":"none","parameter_count":0,"one_line_summary":"AHORN is a standalone, versioned repository of 83 standardized higher-order network datasets with web, API, and Python access.","lead":"The authors built AHORN, a curated online repository that standardizes and version-publishes 83 higher-order network datasets, with a web catalog, a JSON API, and Python tools. It gives network scientists a citable, searchable, reproducible way to find and reuse datasets of group interactions.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Central claim depends on an unverified live service: the 83-dataset/94-revision inventory and the associated access URLs cannot be checked from the manuscript alone.","rationale":"The reader's weakest assumption identifies precisely the same load-bearing concern: the central claim is about a functioning, citable, versioned repository, but the manuscript alone does not establish that the described platform and inventory currently exist. My stress-test found no stronger internal objection. The paper is transparent about limitations, provides an open-source workflow, and describes reproducible converters, but those are design claims; the quantitative coverage claims (83 datasets, 94 revisions, statistics in Figures 2–4) and the access routes (Section 6) are all external, time-sensitive facts. Since the reader already recommended conditional acceptance pending live verification, my read does not change the verdict. The concrete test would settle the concern by checking the live API, Zenodo records, and a sample of downloadable datasets.","tokens_in":6181,"tokens_out":3022,"duration_ms":34025,"concrete_test":"Fetch https://ahorn.rwth-aachen.de/api/datasets.json, parse the JSON, and count the dataset identifiers and revisions. Independently query the Zenodo community https://zenodo.org/communities/ahorn/ for records/revisions. Download 3–5 datasets via ahorn-loader with revision pinning, run the validator, and confirm HIF exports open in XGI or another HIF-compatible library. Compare against Section 4's '83 datasets with 94 revisions' and the structural statistics (e.g., 75 hypergraphs, 44 simplicial complexes). If the counts differ, URLs fail, or exports cannot be read, Sections 4 and 6 are unsupported.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's central claim (Section 2.1) is that AHORN is a functioning repository that transforms public sources into standardized higher-order artifacts, documents provenance, and provides access. The support for this claim consists of the inventory counts in Section 4 ('83 datasets with 94 revisions'), the structural statistics, and the URLs in Section 6. None of this is independently checkable from the manuscript text: the counts are author-reported, the 'complete dataset catalog' is not included in the visible text, and the URLs resolve to mutable live services. A mismatch, outage, or changed inventory would not merely weaken a peripheral remark—it would undermine the central claim that the repository exists with the described coverage and access. This is an external-verifiability concern rather than an internal inconsistency: the design narrative could be accurate while the live artifact is unavailable or different. The paper provides no snapshot hash, revision list, or archived copy of the catalog that would let a referee falsify the claims without network access. Therefore the central claim is conditional on live verification.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","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.","tokens_in":6434,"tokens_out":5252,"duration_ms":54055,"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":[{"comment":"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.","section":"§4, §6"},{"comment":"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.","section":"§4, §5"}],"minor_comments":[{"comment":"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.","section":"§2.2"},{"comment":"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.","section":"Table 1"},{"comment":"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.","section":"§1.1"},{"comment":"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.","section":"Figure 4"},{"comment":"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.","section":"§5"}],"recommendation":"major_revision","confidential_remarks":"This is a data-infrastructure paper; its value depends on the actual availability and correctness of the described repository. If the live URLs and the supplementary catalog are independently verified by the editor, the remaining concerns are minor. I would recommend asking the authors to provide a static snapshot or checksummed catalog and to clarify the parent/child dataset count before acceptance."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague — this is a solid infrastructure paper. The authors built a real thing: a standalone higher-order network dataset repository, not bolted onto a library, with versioned Zenodo releases, a machine-readable API, a validation tool, and HIF exports. That combination is new relative to XGI-DATA and hypergraphx-data, which are tied to their ecosystems. The paper describes the architecture, workflow, and catalog statistics clearly and is internally consistent. It also honestly lists limitations: domain coverage, metadata incompleteness, and the treatment of derived datasets as independent observations.\n\nThe main soft spot is the external verifiability of the central claim. The inventory counts (83 datasets, 94 revisions) and the structural statistics are self-reported, and the full catalog is supposed to be in supplementary material but is not included in the version I saw. The URLs point to a live service, so they can be checked right now, but they could also go down or change between review and publication. The stress test is right: the paper's core assertion that AHORN exists with this coverage and access needs live confirmation. That's not a fatal flaw for an infrastructure paper; I'd just make the snapshot available in a citable form at submission time. A static catalog file, or a commit hash for the dataset list, would settle it.\n\nA related, smaller point: the paper claims the validator and tests exist, but the evidence is the code link. That's fine, but the paper itself doesn't demonstrate a validation run on all 83 datasets. Not a big deal; I'd just want the CI badge or a script output in the repo.\n\nThe citation pattern looks fair. They cite the prior repositories and the reference collection by Benson, and they don't overstate their novelty. The integration with TopoNetX is plainly stated.\n\nWho's this for: researchers working with higher-order network datasets who want standardized, citable data, and anyone building similar community infrastructure. The paper deserves a serious referee. I'd recommend conditional acceptance with the request to make the catalog snapshot part of the article package and to confirm the URLs at proof stage.","headline":"A useful, clearly written infrastructure paper that needs its live repository snapshot pinned down before acceptance.","tokens_in":6846,"tokens_out":2106,"would_cite":true,"duration_ms":23240,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper introduces AHORN, a curated repository that standardizes 83 higher-order network datasets and documents each one's provenance.","keywords":["higher-order networks","hypergraphs","simplicial complexes","data repository","dataset provenance","data curation","FAIR data","benchmark datasets"],"falsifier":"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.","tokens_in":6131,"feed_emoji":"🕸️","tokens_out":7656,"duration_ms":77752,"temperature":0.7,"pith_summary":"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.","feed_headline":"New repository standardizes 83 higher-order network datasets","feed_subtitle":"Each dataset links to its source, citation guidance, conversion code, and fixed version snapshots for reproducible reuse.","key_machinery":"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","core_discovery":"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","pith_inferences":["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."],"forward_implications":["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."],"supporting_citations":[{"why":"The related higher-order hypergraph repository that AHORN positions against; its reported URL was unavailable when checked.","marker":"[7]"},{"why":"The software ecosystem whose bundled data resource is a related, tool-specific effort used as a comparison point.","marker":"[8]"},{"why":"The upstream collection that supplies most of AHORN's initial source datasets, including coauthorship, legislative, contact, and review data.","marker":"[11]"},{"why":"A mature general-purpose graph repository in the comparison table that AHORN distinguishes its higher-order focus from.","marker":"[12]"},{"why":"The manifold-triangulation benchmark family that supplies geometric source datasets and exercises converters on matrix-based inputs.","marker":"[17]"},{"why":"The datasheet framework that loosely inspires the provenance and metadata structure of each dataset page.","marker":"[18]"},{"why":"The data-citation principles that motivate releasing versioned, citable dataset revisions.","marker":"[19]"},{"why":"The interchange format that AHORN also exports to, enabling reuse by compatible analysis libraries.","marker":"[20]"}],"fun_headline_variants":["AHORN: a repository standardizing 83 higher-order network datasets","83 higher-order network datasets, now standardized and versioned","New repo makes 83 higher-order network datasets reproducible and citeable","AHORN curates 83 higher-order datasets with full provenance","Versioned higher-order network datasets: AHORN's 83-set catalog"],"cache_read_input_tokens":2688,"weakest_assumption_plain":"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.","fun_headline_variants_meta":{"raw":{"variants":["AHORN: a repository standardizing 83 higher-order network datasets","83 higher-order network datasets, now standardized and versioned","New repo makes 83 higher-order network datasets reproducible and citeable","AHORN curates 83 higher-order datasets with full provenance","Versioned higher-order network datasets: AHORN's 83-set catalog"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000335,"raw_usage":{"total_tokens":1626,"prompt_tokens":610,"completion_tokens":1016,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":354,"completion_tokens_details":{"reasoning_tokens":927}},"tokens_in":354,"tokens_out":1016,"duration_ms":10646,"temperature":1.0,"reasoning_tokens":927,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-05T17:58:46.636073+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"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.","supporting_citations":[{"cited_title":"& Spagnuolo, C","cited_arxiv_id":null,"evidence_quote":"The related higher-order hypergraph repository that AHORN positions against; its reported URL was unavailable when checked."},{"cited_title":"W.et al.XGI: A Python package for higher-order interaction networks.Journal of Open Source Software8, 5162 (2023)","cited_arxiv_id":null,"evidence_quote":"The software ecosystem whose bundled data resource is a related, tool-specific effort used as a comparison point."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"The upstream collection that supplies most of AHORN's initial source datasets, including coauthorship, legislative, contact, and review data."},{"cited_title":"KONECT: The Koblenz network collection","cited_arxiv_id":null,"evidence_quote":"A mature general-purpose graph repository in the comparison table that AHORN distinguishes its higher-order focus from."},{"cited_title":"InThe Thirteenth International Conference on Learning Representations(2025)","cited_arxiv_id":null,"evidence_quote":"The manifold-triangulation benchmark family that supplies geometric source datasets and exercises converters on matrix-based inputs."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"The datasheet framework that loosely inspires the provenance and metadata structure of each dataset page."},{"cited_title":"Joint Declaration of Data Citation Principles","cited_arxiv_id":null,"evidence_quote":"The data-citation principles that motivate releasing versioned, citable dataset revisions."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"The interchange format that AHORN also exports to, enabling reuse by compatible analysis libraries."}],"review_version":1}