ModelKG : A Knowledge Graph of Plant Photosynthesis and Hydraulic Model Parameters

ModelKG is a machine-readable dataset of structured records extracted from the peer-reviewed literature on plant photosynthesis, stomatal conductance, and plant hydraulic models published between 1986 and 2026. The dataset links research papers to model mentions, mathematical equations, and numerical model parameters. Records include provenance and quality-related fields such as DOI, source context, page number where available, extraction source, confidence, quality flags, species, and experimental conditions. The deposit contains the following files: - model_mentions.jsonl: 44,510 paper–model association records.- equations.jsonl: 3,603 equation records, including LaTeX and plain-text representations.- parameters.jsonl: 29,721 parameter records, including Vcmax, Jmax, gs, gm, Rd, ETR, and other photosynthesis- and hydraulics-related variables.- parameters_warn.jsonl: 1,705 suspicious-tier parameter records retained for quality-control analyses.- parameters_rejected.jsonl: 834 rejected records that can be used as a negative set for quality-assurance workflows.- paper_metadata.jsonl: 12,368 DOI-level records containing species and experimental-condition information.- document_provenance.csv: 17,047 bibliographic provenance records for the processed source documents. The dataset is intended to support analyses of photosynthesis and hydraulic model parameters, knowledge-graph construction, literature-based retrieval, and reproducibility workflows. The released records can also be used to identify parameter values and model representations for downstream comparative analyses. The deposit does not include publisher full-text PDF or Markdown files, raw OCR files, pre-deduplication intermediates, or Neo4j and ChromaDB binary stores. The document_provenance.csv file contains bibliographic metadata only. Short context fields are retained in selected records to help users trace extracted values to their source passages. The MANIFEST.tsv file lists the released files, record counts, and SHA-256 checksums. CHECKSUMS.sha256 provides checksums for the seven data files. The released dataset is distributed under the Creative Commons Attribution 4.0 International license (CC BY 4.0). Users should cite this Zenodo record using its DOI and consult the source-paper DOIs included in the records when relying on specific measurements or source-derived information.