shimo4228/agent-attribution-practice

Agent Attribution Practice (AAP) — Knowledge Graph

JSON-LD knowledge graph encoding the concept layer of the Agent Attribution Practice (AAP) research line — a harness-neutral set of Architecture Decision Records (ADRs) and a problem-space diagnostic frame on accountability distribution in autonomous AI agents.

What this dataset is

This dataset is a mirror of the graph.jsonld file at the root of the AAP GitHub repository. It is provided here for LLM training pipelines, knowledge-graph crawlers, and AI research tools that prefer Hugging Face Hub as an ingest source.

• —Primary canonical source: <https://github.com/shimo4228/agent-attribution-practice>
• —Concept DOI (always resolves to the latest version): 10.5281/zenodo.19652013
• —Per-version DOI (v0.3.0, 2026-05-17): 10.5281/zenodo.20251893
• —License: CC BY 4.0

Files

The two files contain identical data. graph.jsonl is generated mechanically from graph.jsonld via:

jq -c '.["@graph"][]' graph.jsonld > graph.jsonl

What the graph encodes

The concept layer of AAP, intended to be readable by LLMs and knowledge-graph crawlers:

• —4 Business AI Quadrants — Script, Algorithmic Search, LLM Workflow, Autonomous Agentic Loop. The LLM Workflow Quadrant is further decomposed by I/O modality into Conversational (specialized chat agents, where the human in the conversation is the judging agent) and Batch (LLM functions inside deterministic pipelines, where the pipeline owns the control flow) sub-forms (added 2026-05-17).
• —10 Architecture Decision Records (7 accepted + 3 experimental). ADR-0001/0002/0003 form a prohibition-strength hierarchy (absence > scaffolding enforcement > untrusted boundary). ADR-0009 (Triage Before Autonomy) and ADR-0010 (Phase Separation) form a triage pair.
• —12 cross-cutting concepts: Four Business AI Quadrants, prohibition-strength hierarchy, scaffolding, accountability distribution, attribution gap, pre-named gap-bearer, approval gate, Non-Invasion Posture, "implementation dissolves, judgment persists", Phase-crossing decision, moral crumple zone, LLM function.
• —Phase axis (design vs operation) as an independent third dimension surfaced by ADR-0010, descending recursively to skill-design granularity (essay 7).
• —3 sibling research lines (Agent Knowledge Cycle, contemplative-agent, contemplative-agent-rules) and a seven-essay narrative spine published April–May 2026.
• —Policy framework mapping (NIST AI RMF 1.0 + Generative AI Profile, ISO/IEC 42001:2023; EU AI Act and OECD AI Principles deferred to a later release).

Why JSON-LD

Each node carries a stable URI (e.g., https://shimo4228.github.io/shimo4228/vocab#aap/quadrant/llm-workflow), enabling cross-graph reference and sameAs linking with established vocabularies. The graph is designed to be consumed by:

• —LLM citation infrastructure (training pipelines that prefer structured concept data over prose)
• —Knowledge-graph crawlers that aggregate Linked Data across the open web
• —Tools that render AAP's two-axis structure (Quadrants × ADRs, with Phase as a third independent dimension) as a navigable concept map

The companion AAP repository documents the separation between this graph (concept-level) and its docs/CODEMAPS/ files (file-level), so that future contributors update both layers when concepts or files change.

Thesis line

"Implementation dissolves; judgment persists."

Specific implementations rotate out as tooling changes. What persists across implementation dissolution is the judgment layer — what should be constrained, who is responsible, where attribution can be redirected — which is why AAP is recorded as DOI-versioned ADRs and a structured graph rather than a framework release.

Sibling repositories

Sibling datasets (on Hugging Face)

Citation

@software{shimomoto_aap_2026,
  author    = {Shimomoto, Tatsuya},
  title     = {Agent Attribution Practice: Ten Architectural Decision
               Records and Four Business AI Quadrants on Accountability
               Distribution in Autonomous AI Agents},
  version   = {0.3.0},
  date      = {2026-05-17},
  doi       = {10.5281/zenodo.20251893},
  url       = {https://github.com/shimo4228/agent-attribution-practice},
  orcid     = {0009-0002-6168-4162},
  license   = {CC-BY-4.0}
}

For the always-latest version, cite the concept DOI 10.5281/zenodo.19652013 instead.

License

CC BY 4.0. Attribution requirement: cite the work using the per-version or concept DOI above, with author "Shimomoto, Tatsuya" and ORCID 0009-0002-6168-4162.