The zbMATH Open Knowledge Graph: Tracing Centuries of Mathematical Research
cs.DL · cs.AI
Quick verdict
Claims a 34M-entity, 168M-triple KG of 250+ years of math scholarship; a solid open resource, but its utility is shown by anecdote, not systematic evaluation.
Summary
The paper presents the zbMATH Open Knowledge Graph, an RDF knowledge graph built from a December 2025 snapshot of the zbMATH Open platform API. It integrates more than 250 years of mathematical scholarship — 4,070,393 articles spanning 1763–2026 — with expert-curated semantic content: 3,098,767 expert-written reviews, 3,008,881 controlled keywords, 6,734 MSC classification concepts, 30,857 software references, and 1,125,474 disambiguated persons, totaling 34.1 million entities and 168.7 million RDF triples modeled with schema.org, Dublin Core, CiTO, and SKOS. Construction follows METHONTOLOGY with SAMOD and LOT practices; data, ontology, code, and a public SPARQL endpoint are released under CC-BY-SA 4.0. Evaluation is three-tiered: syntactic validation (Jena, RDFLib; loading into Virtuoso and Fuseki), SPARQL structural checks that surfaced 61,380 missing-property violations (61,377 missing authors, traced to incomplete upstream API records), and 45 competency questions, of which 86.7% are fully and 95.6% at least partially supported. Five 'historically grounded scholarly exploration' use cases illustrate precursor discovery, cross-field concept continuity, conceptual resurgence, and reviewer–author connections. Main limitations: use-case utility is demonstrated anecdotally, without precision estimates or baselines; the competency-question evaluation is a manual, single-rater exercise; and Table 1's entity/resource metrics are ambiguously defined.
Review complete
Assessment at a glance
4
risk flags identified
● high 0
● medium 2
● low 2
5 strengths · 4 weaknesses · 5 author questions
AI text signal = Binoculars zero-shot detector score over 20 sampled passages (Fast-DetectGPT co-reported) — absence does not prove human authorship, and a flag does not imply misconduct.
Levels reflect verification complexity — not a quality score. Evolve never rates your paper.
Key claims under review
- The zbMATH Open KG couples 250+ years of mathematical bibliographic data (4,070,393 articles, 1763–2026) with expert-curated semantics — reviews, controlled keywords, MSC codes, software references, disambiguated authorship — at a scale (34.1M entities, 168.7M triples) not offered by citation-centric scholarly KGs.
- The graph is syntactically valid, structurally consistent apart from 61,380 violations traceable to incomplete upstream records, and fully supports 86.7% of 45 competency questions.
- Combining curated semantics with deep temporal coverage enables 'historically grounded scholarly exploration' (precursor discovery, cross-field continuity, resurgence, reviewer–author links) that bibliographic metadata and citation networks alone cannot support.
- The resource is FAIR-aligned and sustainable: CC-BY-SA 4.0, versioned semiannual Zenodo dumps, public SPARQL endpoint, persistent URIs, and a released reproducible construction pipeline.
Strengths
Unique expert-curated content at scale
Tables 1–2: 34,099,406 entities / 168,670,821 triples combining 4,070,393 articles (1763–2026) with 3,098,767 expert reviews, 3,008,881 controlled keywords, 6,734 MSC concepts, 30,857 software references and 1,125,474 disambiguated persons — content absent from citation-centric scholarly KGs (§2).
Standards-based, fully open release
§3.2/§5: systematic reuse of schema.org, dcterms, CiTO and SKOS; VoID/DCAT/PROV-O metadata; versioned Zenodo dumps (~2 releases/year), public SPARQL endpoint, URI resolution, and full pipeline code under CC-BY-SA 4.0.
Honest failure reporting
Reproducible three-tier evaluation
Disciplined hedging in use cases
Weaknesses
Anecdotal utility evaluation
Single-rater manual CQ evaluation
Inconsistent Table 1 metric definitions
Headline disambiguation and citation claims unevaluated
Full Review Report
$1.99 · one-timeUnlock the complete preliminary analysis Evolve produced for this paper.
Report preview
Scope & Claims
This is a resource/dataset paper (arXiv 2609.00969, cs.DL/cs.AI) presenting the **zbMATH Open Knowledge Graph**, an RDF KG derived from the zbMATH Open platform's API (December 2025 snapshot). Central claims: (1) the KG uniquely couples bibliographic depth — 4,070,393 scholarly articles spanning 1763–2026 (Table 2) — with expert-curated semantics: 3,098,767 reviews, 3,008,881 controlled keywords, 6,734 MSC concepts, 30,857 software references, and 1,125,474 disambiguated persons, totaling 34,099,406 entities and 168,670,821 triples (Table 1); (2) it is standards-based (schema.org, dcterms, CiTO, SKOS) and FAIR-aligned; (3) it enables 'historically grounded scholarly exploration' beyond citation networks. The resource ships under CC-BY-SA 4.0 with Zenodo dumps, a public SPARQL endpoint, and full pipeline code.
The remaining sections cover Methodology Assessment, Statistical Evidence, Results & Reproducibility, Limitations & Risks, and Verdict & Recommendations.
- All 5 strengths & 4 weaknesses, analyzed in detail
- 4 risk flags with severity, section & reviewer questions
- Methods & statistical evidence assessment
- 5 questions a peer reviewer would ask
- Full six-dimension evaluation report & recommendations
One-time purchase · this paper only
One-time purchase, this paper only. AI-generated preliminary analysis — not peer review.
Frequently asked questions
What is the zbMATH Open Knowledge Graph?
An open RDF knowledge graph built by FIZ Karlsruhe from the zbMATH Open mathematics reviewing service. It covers 1763–2026 with about 4.07 million articles, 3.1 million expert-written reviews, 3 million controlled keywords, 6,734 MSC classification concepts, and 1.13 million disambiguated scholars — 34.1 million entities and 168.7 million RDF triples in total.
How can I access or query the zbMATH Open Knowledge Graph?
Two ways: download versioned RDF dumps from Zenodo (CC-BY-SA 4.0 license) and load them into any triple store (tested with Virtuoso and Jena Fuseki), or query the public SPARQL endpoint linked from the GitHub repository; core entity URIs also resolve as linked data.
What does the zbMATH Open KG offer beyond citation databases like OpenAlex or Semantic Scholar?
Expert-curated semantics: reviews, controlled keywords, and MSC subject codes let users link works by shared concepts even when no citation exists — for example finding potential precursor papers or tracking a concept across mathematical fields. The paper demonstrates this on examples but does not yet quantify how accurate such discovered connections are.
About this review
This is an AI-generated preliminary review by Evolve.Pub, produced from the openly available text of the paper. It is not peer review and does not represent the authors' views. The canonical record of the paper is at the publisher link above.
- The five figures render as garbled embedded graphics in the extracted text, so figure-internal details (e.g., exact decade-level counts in Fig. 4) could not be verified; figure captions and in-text descriptions were used instead. All cited numbers above come from tables and body text.
Go deeper on this paper
Wrote a paper? Stress-test it before reviewers do.
Upload your manuscript for the same structured scrutiny — reasoning, statistics, evidence, figures, and reproducibility.