Lincoln-Rwodzi/ahodo-discovery
AHODO Discovery Dataset v0.3 AHODO is a cross-institutional discovery and rights/provenance metadata dataset for African humanities and humanities-adjacent resources. This v0.3 distribution contains 11,650 records. It is a discovery registry, not a corpus of the works it describes or a representative sample of African humanities. It is not presented as an AI-training dataset. Interactive search Canonical Zenodo archive and DOI Zenodo record Public GitHub repository… See the full description on the dataset page: https://huggingface.co/datasets/Lincoln-Rwodzi/ahodo-discovery.
AHODO Discovery Dataset v0.3
AHODO is a cross-institutional discovery and rights/provenance metadata dataset for African humanities and humanities-adjacent resources. This v0.3 distribution contains 11,650 records. It is a discovery registry, not a corpus of the works it describes or a representative sample of African humanities. It is not presented as an AI-training dataset.
Purpose and scope
AHODO asks: What African humanities data exists, where is it held, and under what legal, ethical and community-defined conditions can it be reused? It supports discovery across repositories, museums, publishers and periodicals. Records keep holding location distinct from subject geography, access distinct from rights, and metadata/component rights distinct from rights in underlying works. Copyright/licence findings, computational-reuse evidence, AI-specific terms, other rights, community-governance information and provenance are separate assessments.
Important: discovery does not equal permission
AHODO does not redistribute the underlying cultural or scholarly works represented by its records. Public access does not establish an open licence. Inclusion in AHODO is not blanket permission for reproduction, redistribution, computational reuse, AI/ML training or commercial use. Rights may differ among metadata, catalogue text, images, full text, audio and other components. Consult each record's rights assessment and the authoritative source institution before reuse.
not_established is not permission. community_governance: not_provided is not evidence that community protocols, interests or restrictions are absent, or that consent was given. A documented copyright/licence basis for a specified computational operation is not blanket AI permission. The Zenodo DOI provides citation and persistence; it grants no additional rights. See rights and provenance, the repository notice, licensing guidance and the public source manifest.
Composition
The released records come from 12 source families and 12 holding institutions. Source/acquisition categories include institutional repositories, museum APIs, publisher catalogue pages and a journal archive; the public source manifest documents their provenance. Holding country is not subject geography: an item held in one country may concern another. AHODO does not infer subject geography from institutional location or language.
Language metadata is incomplete: 4,950 records have no stated language. An absent language code does not show that the work is not in that language. Language is not inferred from titles. In this release, no active item asserts Shona as its recorded language; textual relevance to Shona is a different claim. The language: en card tag describes the primary language of this dataset's documentation/metadata, not the languages of all represented resources.
Data and fields
data/ahodo-discovery-v0.3-complete.jsonl is the single complete structured export. The Hub configuration exposes it as one neutral data split; there are no train, test or validation partitions. Each physical line is one JSON object. The file is the corrected, byte-identical JSONL from the published v0.3 Zenodo package.
Main fields include id, source_id, authoritative_id, canonical_url and its status, title, material_types, contributors, dates, languages, subject_geography, holding_institution, scope_classification, rights_reference, rights_assessment, community_governance and public_provenance. Optional licensed_catalogue_text appears where supplied. See the data dictionary for field meanings.
The Rights Passport approach assesses rights by exact resource/component where evidence permits. A licence for a catalogue-text component does not automatically apply to an artwork, image, linked PDF or full text. The dataset separately records access, copyright/licence, computational-reuse basis, source AI-specific statements or restrictions, other rights and community governance where reviewed. The repository-level licensing notice does not override source-derived metadata licences, CC0 components, third-party rights or rights in underlying works.
Appropriate uses
The dataset supports humanities resource discovery, cross-institutional discovery, repository landscape analysis, provenance research, rights-metadata research, cultural heritage metadata analysis, language metadata analysis and identifying records requiring further rights review. Researchers considering computational analysis must determine whether their intended operation is permitted for the relevant data and components, including applicable non-copyright conditions.
Limitations
Coverage is incomplete and uneven across institutions. The largest source contributes 3,461 records. Subject geography is not_established for 11,496 records; recorded language is missing for 4,950. Rights findings depend on available evidence: 10,738 records reference metadata without an established licence. The absence of recorded community-governance information is not evidence of its absence. Source terminology varies, and metadata-based scope or possible-duplicate assessments can be uncertain. AHODO does not replace authoritative repositories; their records and terms may change after this frozen release. See known limitations.
Citation and versioning
The canonical archived snapshot is AHODO Discovery Dataset v0.3: African Humanities Resource Discovery and Rights Metadata, DOI: 10.5281/zenodo.22944185. Cite: Rwodzi, L. T., & Gotosa, K. (2026, September 25). AHODO Discovery Dataset v0.3: African Humanities Resource Discovery and Rights Metadata [Dataset]. Version 0.3. https://doi.org/10.5281/zenodo.22944185. Cite relevant source institutions and resources separately; see CITATION.cff.
This Hugging Face copy distributes v0.3 for inspection and access. AHODO Search is the interactive interface, and GitHub hosts public documentation and schema. Future dataset versions should have their own archival release and versioning process; v0.3 records should not be silently replaced.