augmanitai

AUGMANITAI

Explore the public work

AI-generated navigation guide, added 21 September 2026.

Deutsch: Einstieg in die öffentlichen Arbeiten von Andreas Ehstand: Themenübersicht, Begriffe, konzeptionelle Texte und kreative Arbeiten. Angaben zu Quellen, Lizenzen und Grenzen stehen bei den jeweiligen Materialien.

A terminology framework for human–AI interaction. 1000 curated terms, informed by ISO 704 / ISO 1087 / ISO 30042 (ISO-inspired, not ISO-certified), Semantic Web native.

DOI License: CC BY-NC-ND 4.0 License: Apache 2.0 ORCID

Author: Andreas Ehstand Research programme concept DOI: 10.5281/zenodo.20161494 Status: Priority-establishing terminological deposit. The 1000-Term Compendium is the curated output of a longitudinal n=1 observation of daily human–LLM interaction over approximately nine months, published as a timestamped Zenodo record with Semantic Web format extensions, validation shapes, and reference infrastructure. This is not a peer-reviewed publication and does not target one; see METHODOLOGY.md §2.0 and §11 for the design rationale.

⚠️ Ethical Disclaimer (§1–§20) — please read before use

AUGMANITAI is purely descriptive linguistic-terminological research into phenomena observable at the human–AI interface. It is not normative, makes no recommendations for action, and is not a diagnostic, medical, psychological, legal, or screening instrument.

The full bilingual §1–§20 disclaimer is in DISCLAIMER.md and applies to every file, every term, and every serialization in this repository.

The disclaimer is complemented by two transparency documents that every reviewer, replicator, or critic is expected to read:

Fictional examples only. Where definitions in this repository are illustrated by examples of human–LLM interaction, the examples are either fully synthetic or de-identified paraphrases of public discussion. No private chat logs, correspondence, or re-identifiable third parties appear in any term, definition, example, or serialization. See METHODOLOGY.md §5.

Note on the word “phenomenology”. Where this repository uses the word phenomenology or phenomenal, it means observable regularities that can be named and described. It does not invoke the Husserlian, Merleau-Pontian, or broader continental-philosophical tradition of phenomenology. No claims from that tradition are transferred to AUGMANITAI. See METHODOLOGY.md §7.

License: CC BY-NC-ND 4.0 (data) · Apache-2.0 (code). Jurisdiction: Munich, Germany.

What AUGMANITAI is

AUGMANITAI is a structured terminology for phenomena that arise at the interface between humans and large language models. The core deliverable is the 1000-Term Compendium: a curated set of terms, each with definition, genus/differentia structure, characteristics, and relations — published under a permanent DOI and formalized as an RDF knowledge graph.

The compendium is the curated output of a longitudinal n=1 observation conducted by a single observer over approximately nine months of daily, extended human–LLM interaction. It is a timestamped case record, comparable in form to a longitudinal case report or a qualitative single-subject study, transposed onto the human–LLM interaction domain.

What is claimed and what is not. AUGMANITAI v1.0 claims priority on the schema as a whole — the selection, definition, relation structure, and W3C-native serialization, deposited on Zenodo with a cryptographically anchored timestamp. It does not claim invention of every individual term: some are drawn from existing literature, some are re-formulations of terms already in use, and some are labels newly introduced by the curator where no existing term fit. The three origin categories are described in METHODOLOGY.md §3. A reader who identifies prior art for any term is welcome — the correct response is attribution, not rebuttal.

The framework is intended as a research instrument. It is not a product, not a methodology, and not a manifesto. Terms are tools for observation.

What is in this repository

Directory Contents
ontology/ Core OWL ontology, SKOS concept scheme, OntoLex-Lemon, SKOS-XL, JSON-LD context, SHACL shapes, TBX export
protocols/ MCP, A2A, ANP, and AG-UI protocol adapters for agent integration
provenance/ C2PA manifests, AI-SBOM (SPDX 3.0 + CycloneDX 1.6 ML-BOM), W3C Verifiable Credentials 2.0
well-known/ llms.txt — LLM discoverability pointer to the canonical Zenodo DOI
graphrag-variants/ GraphRAG integration examples (Microsoft, LlamaIndex, LangGraph, HippoRAG, LazyGraphRAG, FalkorDB)
demo-swarms/ Reference multi-agent examples (LangGraph, CrewAI, AutoGen) with mock LLMs
governance/ Code of Conduct, CLA, DCO, Security, Maintainers, Roadmap, Versioning
licensing/ Dual-license files (CC BY-NC-ND 4.0 for data, Apache 2.0 for code), SPDX, PATENT, TRADEMARK, NOTICE

Quick start

Load the ontology

# Apache Jena
riot --validate ontology/augmanitai-ontology.ttl

# rdflib (Python)
python3 -c "import rdflib; g=rdflib.Graph(); g.parse('ontology/augmanitai-ontology.ttl'); print(len(g), 'triples')"

Validate data with SHACL

pip install pyshacl
pyshacl -s ontology/validation/shacl-shapes/term-shapes.ttl \
        -d ontology/augmanitai-ontology.ttl

Explore with GraphRAG

See graphrag-variants/ for six integration examples with mock LLMs that run offline.

Standards the framework is informed by

Per §16 of the Ethical Disclaimer, AUGMANITAI is ISO-inspired, not ISO-certified. The table below lists the standards whose principles informed the methodology; it does not assert certification, conformance, or official approval by any standards body.

Standard Role
ISO 704:2022 Terminology work — principles and methods (informed by)
ISO 1087:2019 Terminology work — vocabulary (informed by)
ISO 30042:2019 TermBase eXchange (TBX) serialization (informed by)
W3C SKOS Concept scheme
W3C SKOS-XL Extended labels with per-label provenance
W3C OntoLex-Lemon Lexical layer
W3C SHACL Validation shapes
W3C JSON-LD 1.1 Web-native serialization
W3C Verifiable Credentials 2.0 Content provenance
SPDX 3.0 AI Profile Bill of materials
C2PA 2.0 Content credentials

Licensing

AUGMANITAI uses a dual-license model. The two licenses cover disjoint sets of files — there are no mixed-license files in this repository.

Data layer — CC BY-NC-ND 4.0:

Code layer — Apache License 2.0:

SPDX expression: CC-BY-NC-ND-4.0 AND Apache-2.0

No mixed works. The repository deliberately avoids files that combine terminological content (which is data and non-derivable) with executable logic (which is code and permissively licensable). A Python script that loads an ontology is Apache-2.0; the ontology it loads is CC BY-NC-ND 4.0. A demo that embeds term definitions inline must either split into two files or be treated as a data file.

Contribution implications.

See licensing/ for the full policy, patent non-assertion, trademark guidance, and third-party attributions.

For commercial licensing inquiries regarding the data layer, contact the author via ORCID.

Citation

The concept DOI below currently resolves to the research programme master index. It does not identify a fixed version of the 1000-Term Compendium; citations to a specific compendium release should use that release’s own record.

@article{ehstand_augmanitai_programme,
  author       = {Ehstand, Andreas},
  title        = {AUGMANITAI Research Programme — Concept-DOI Master Index V7: NEOMANITAI Methodology Cluster (2026), by Andreas Ehstand},
  publisher    = {Zenodo},
  doi          = {10.5281/zenodo.20161494},
  url          = {https://doi.org/10.5281/zenodo.20161494}
}

A machine-readable citation is available in CITATION.cff.

Contributing

AUGMANITAI is curated terminology work, not an open commons of editable definitions. All contributors acknowledge and accept the Ethical Disclaimer (DISCLAIMER.md, §1–§20) as a precondition of contribution. Contributions are welcome but governed:

All contributors agree to the Code of Conduct.

See CONTRIBUTING.md for details.

Security

Vulnerabilities and security concerns: see SECURITY.md. Do not report security issues via public GitHub issues.

Governance

Acknowledgements

Built on the shoulders of ISO, W3C, DCMI, PROV-O, schema.org, SPDX, C2PA, Wikidata, the MCP/A2A/ANP/AG-UI communities, and the Contributor Covenant. Full attributions in licensing/NOTICE.

Ecosystem

AUGMANITAI is the foundation of the MANITAI framework ecosystem:

This site is open source. Improve this page.