myco-spore-nutrients

# MycoNet NutrientCards — Derived Dataset / 菌网养分卡·衍生数据集 **EN.** A derived-only dataset of MycoNet NutrientCards (Myco-Spore factory). Each line in `data/nutrients.jsonl` carries the *derived* layer of a source: a structured summary, a derived meaning, a decomposition proof, and provenance / attribution. The **full source text is intentionally NOT included** — only licensed-derived artifacts are redistributed, keeping the dataset safe to share across borders. **中文.** 本数据集仅包含菌网养分卡（Myco-Spore 工厂）的**衍生层**：结构化摘要、 衍生语义、分解证明，以及溯源 / 署名信息。**不包含任何源文全文**——只再分发经许可 派生的内容，确保跨境分发合规安全。 ## License / 许可 - The **derived content** (summaries, derived meaning, proofs) is released under **CC-BY 4.0** — see `LICENSE`. - The **original sources** retain their own licenses (`license_class`): `public-domain` (e.g. Project Gutenberg) or `open-license`. Attribution to the source is preserved via `source_url` and `attribution`. ## Fields / 字段 15 fields: card_id, source_url, status, license_class, lang, structured_summary, derived_meaning, derived_confidence, decomposition_proof, snapshot_cid, attribution, embedding_ref, domain, domain_confidence, origin. See `schema.json` for the machine-readable schema. ## Domains / 域分布 (SPEC-006b) Coarse controlled-domain label per card (`domain`). Distribution across this release: - **mycology**: 815 - **ecology**: 43 - **biology**: 300 - **medicine**: 740 - **literature**: 51 - **general**: 614 Discovery tags (non-empty domains): mycology, ecology, biology, medicine, literature ## Origin / 来源纯度法 (SPEC-007) Purity-law split per card (`origin`): - **living**: healthy, active, currently-living sources (factory whitelist + live sitemap hosts). - **garbage**: real network refuse — dead links, archives, rot recovered from the wild. - **garbage**: 0 - **living**: 2563