Datasets:
$schema string | $id string | title string | description string | type string | required list | additionalProperties bool | properties dict |
|---|---|---|---|---|---|---|---|
https://json-schema.org/draft/2020-12/schema | https://github.com/spectralbranding/paper-spec/schema/paper-spec.schema.json | paper-spec | Schema for paper.yaml — a machine-readable specification of a research paper's claims, methodology, and replication requirements. | object | [
"meta"
] | false | {
"meta": {
"description": "Bibliographic metadata for the paper.",
"type": "object",
"required": [
"title",
"authors",
"date"
],
"additionalProperties": false,
"properties": {
"title": {
"type": "string",
"description": "Full title of the paper."
... |
Paper Spec specification corpus
A frozen snapshot of the paper.yaml specification corpus analysed in Machine-readable
specification of the claims, acceptance criteria and dependencies of scientific papers
(Zharnikov, 2026).
Each file specifies one paper's claims, methodology, acceptance criteria, dependencies and limitations in Paper Spec v0.1.0 — a YAML format that sits beside a paper rather than replacing it.
What is here
| path | contents |
|---|---|
examples/ |
27 paper.yaml files, each named for the paper it specifies |
schema/paper-spec.schema.json |
the JSON Schema the files validate against, included so the corpus is interpretable on its own |
LICENSE-data |
CC BY 4.0 |
Why this is a snapshot and not a mirror
The working corpus is live and grows as papers gain specifications — twenty at the date of the study
reported in the paper, twenty-seven here. This deposit is frozen so that a citation points at a
fixed state rather than a moving target. It is taken from
spectralbranding/paper-spec at revision
7fbeb7ff7daf240c88a7d8b3fab21477746717a2.
For the current state of the corpus, the specification document, the reference validator and the per-venue requirement profiles, use the repository. For reproducing the paper's reported figures, use this snapshot.
What the paper computes from it
The authoring times, the dependency findings (the foundational paper a critical dependency for sixteen of the remaining nineteen; a cascade of depth four; two bridge papers), the claim-density range of two to seven claims per paper, the three claims whose falsification conditions were unclear, and the vocabulary result that twenty-seven of thirty vocabulary members have a published predicate already defined for them.
Notes
- The payload is YAML rather than tabular, so the dataset viewer does not render it. That is expected; browse or download the files directly.
- Licence: CC BY 4.0. The executable material in the source repository is MIT and is not included here.
Citation
Zharnikov, D. Paper Spec specification corpus. Hugging Face (2026).
- Downloads last month
- 41