Versioned ontology and vocabulary for the FAIA (FAIR AI Attribution) framework. For background on how to generate a statement from the vocabulary and ontology, see the FAIR AI Attribution framework documentation. To create one directly, use the FAIA statement generator. For other semantic artifacts and integration with the Liccium platform, see Liccium FAIA implementation.
The FAIA framework provides an ontology and vocabulary for machine-readable and persistent disclosure of AI involvement. It defines three complementary elements: high-level attribution flags (human-created, AI-assisted, AI-generated), activity codes describing the role AI played in the content lifecycle, and optional system attribution identifying the AI system and version involved.
FAIA is designed to support regulatory compliance with AI transparency requirements (including the EU AI Act and equivalent legislation) and to enable systematic identification of synthetic content for AI training data governance.
| Artefact | Current version | Status | Canonical file |
|---|---|---|---|
| FAIA ontology | 0.6.0 | final | faia_ont/faia_ont.ttl |
| FAIA vocabulary | 0.7.0 | final | faia_vocab/faia_vocab.ttl |
Each folder also retains earlier versioned snapshots. The unversioned faia_ont.ttl and faia_vocab.ttl files always carry the current release. Persistent version IRIs resolve under https://w3id.org/faia-framework/.
Versioned FAIA vocabulary files in Turtle serialisation. The FAIA vocabulary covers three layers of AI involvement disclosure.
FAIA flags signal the overall level of AI involvement at the content level:
HCC(human-created content): created, generated and edited exclusively by humans or human-controlled instruments; no generative AI at any stage.AAC(AI-assisted content): a human or human-controlled instrument remains the primary creator; AI contributed during the process to various degrees.AIG(AI-generated content): AI is the primary creative agent; human input is limited to prompting, selection, or minor adjustment.
Generic, media-independent codes describing the specific operation AI performed:
Co-Creation: AI and humans jointly produce content through an interactive process; creatorship is shared and not clearly separable.Contribution: AI produces content incorporated into a primarily human-created work; the human remains the final creator of the combined result.Enhancement: AI modifies or extends existing human-created content to improve quality, clarity, richness, or structure.Refinement: AI performs limited adjustments that preserve the existing structure, meaning, and creatorship of the content.Transformation: AI converts existing content into another form or representation while preserving semantic equivalence or a close relationship to the original.Analysis: AI interprets existing content to derive structure, features, metadata, meaning, or insights.
Domain-specific activity codes for academic manuscript preparation, aligned with the STM Association Recommendations for a classification of AI use in academic manuscript preparation:
STM-LanguageRefinement: AI tools suggest or apply refinement, correction, editing, or formatting of manuscript text to improve clarity of language.STM-ManuscriptDrafting: AI tools generate, rewrite, or draft part or all of the manuscript text.STM-ManuscriptTranslation: AI tools assist translation of an author's original work into another language.STM-DataPresentationRefinement: AI tools refine, format, or improve the presentation, clarity, or readability of the data reported in the manuscript or associated materials.STM-IllustrativeFigureGeneration: AI tools generate, refine, correct, edit, or format images, diagrams, or other figures for illustrative or aesthetic purposes only, not as research data or results.STM-ResearchDataVisualisation: AI tools generate, refine, correct, edit, or format visualisations of research data or results for inclusion in the manuscript.STM-CodePresentationRefinement: AI tools refine or format code reported in the manuscript or associated materials to improve readability or clarity, without altering its functionality.STM-ReferenceGathering: AI tools suggest references that may be relevant for inclusion in the manuscript's reference list. Activity codes may also reference further external classifications where appropriate. All flags and activity codes are modelled as SKOS concepts under the FAIA concept scheme.
Versioned FAIA ontology files in Turtle serialisation. The FAIA ontology defines the following data properties on a Declaration:
faiaFlag: indicates AI involvement in the creation or modification of the work, according to the FAIA classification.activityCode: describes the activity or workflow step, aligned with FAIA or STM AI classification.systemAttribution: identifies the AI system or engine used to generate or modify the content, ideally using the provider's official designation.systemVersion: specifies the particular version of the AI system used, following the operator's naming or numbering scheme where known.
The ontology imports the Liccium ontology https://w3id.org/liccium/ont/ for the Declaration class.
These artefacts are maintained by the FAIA framework project team, and are published under the CC BY 4.0 licence. Persistent URLs are managed via w3id.org, operated by the W3C Permanent Identifier Community Group.
For more information:
- FAIA documentation: https://faia.liccium.com
- FAIA registry: https://faia.io
- FAIA statement generator: https://www.faia.io/statement-generator
- Liccium: https://liccium.com