Skip to content
Open
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
114 changes: 79 additions & 35 deletions config/topic_profile.yaml
Original file line number Diff line number Diff line change
@@ -1,9 +1,9 @@
profile_name: ontology_research_agent
main_topic: "Ontology, knowledge graph, semantic web, and AI-assisted literature review / academic writing workflows"
profile_name: ax_ontology_governance_research
main_topic: "Ontology-based AX operating model and generative AI governance, security, and decision automation"

language:
notes: ko
manuscript: en
manuscript: ko
summaries: ko

keywords:
Expand All @@ -13,37 +13,68 @@ keywords:
- knowledge graph
- knowledge graphs
- semantic web
- linked data
- RDF
- OWL
- SHACL
- SPARQL
- SKOS
- taxonomy
- metadata schema
- property graph
- Neo4j
- operational ontology
- enterprise ontology
method:
- literature review automation
- systematic literature review
- research assistant
- academic writing
- scientific writing
- paper recommendation
- citation recommendation
- evidence extraction
- GraphRAG
- retrieval augmented generation
- research agent
- AI scientist
- agentic workflow
- AI agent
- multi-agent system
- decision automation
- decision support system
- workflow automation
- policy-as-code
- rule-based reasoning
- explainable AI
- traceability
- lineage
- auditability
- human-in-the-loop
governance_security:
- generative AI governance
- AI governance
- LLM governance
- AI risk management
- model risk management
- AI security
- LLM security
- prompt injection
- data leakage
- access control
- attribute-based access control
- guardrails
- compliance automation
- trustworthy AI
- responsible AI
application:
- knowledge organization
- scholarly knowledge graph
- research workflow
- document intelligence
- scientific discovery
- AX transformation
- digital transformation
- enterprise AI platform
- AI operating model
- AI operations
- MLOps
- LLMOps
- AIOps
- enterprise architecture
- information systems planning
- public sector AI
- financial AI governance
exclude:
- SEO article
- purely philosophical ontology
- biomedical ontology only
- cryptocurrency ontology
- SEO knowledge graph
- marketing knowledge graph
- purely clinical ontology without reusable methodology
- non-academic blog post
- cryptocurrency ontology
- AI art prompt
- pure chatbot tutorial without governance or evaluation

target_outputs:
- daily_digest
Expand All @@ -53,28 +84,41 @@ target_outputs:
- gap_matrix
- related_work
- manuscript_outline
- governance_framework
- security_control_matrix
- decision_automation_architecture

preferred_venues:
- Semantic Web Journal
- Journal of Web Semantics
- ISWC
- ESWC
- WWW
- The Web Conference
- WWW
- ACL
- EMNLP
- NAACL
- SIGIR
- CHI
- JCDL
- IEEE Access
- ACM Computing Surveys
- Information Systems Frontiers
- Decision Support Systems
- Expert Systems with Applications
- Computers & Security
- IEEE Security & Privacy
- arXiv

query_groups:
- name: ontology_core
query: "ontology knowledge graph semantic web RDF OWL SPARQL"
- name: literature_review_automation
query: "automated literature review research assistant academic writing citation evidence extraction"
- name: scholarly_kg
query: "scholarly knowledge graph scientific literature ontology"
- name: research_agent
query: "AI research agent paper writing scientific discovery literature review"
- name: operational_ontology_ax
query: "operational ontology enterprise knowledge graph AI operating model digital transformation decision automation"
- name: ontology_graphrag_governance
query: "ontology knowledge graph GraphRAG generative AI governance traceability auditability"
- name: llm_governance_security
query: "LLM governance generative AI security prompt injection data leakage guardrails access control"
- name: decision_automation_policy_reasoning
query: "decision automation policy as code rule based reasoning ontology explainable AI human in the loop"
- name: enterprise_ai_operations
query: "enterprise AI platform LLMOps MLOps AI governance risk management compliance automation"
- name: public_financial_ai_governance
query: "public sector AI governance financial AI governance model risk management trustworthy AI"
131 changes: 131 additions & 0 deletions docs/RUN_AX_ONTOLOGY_GOVERNANCE.md
Original file line number Diff line number Diff line change
@@ -0,0 +1,131 @@
# PaperOps Runbook: AX Ontology Governance Topic

## Research topic

**온톨로지 기반 AX 운영체계와 생성형 AI 거버넌스/보안/의사결정 자동화 연구**

Working English topic:

**Ontology-based AX operating model and generative AI governance, security, and decision automation**

## Goal

Use PaperOps to collect, triage, and evidence-govern papers for a master's-level AI/Big Data engineering thesis. The output should support a thesis that treats ontology and knowledge graphs not as general philosophy, but as an engineering mechanism for enterprise AX operations, LLM governance, security control, traceability, and automated decision workflows.

## Success criteria

1. Collect papers across ontology/KG, GraphRAG, LLM governance, AI security, policy-as-code, decision automation, MLOps/LLMOps, and enterprise AI operations.
2. Screen papers into `important`, `to_read`, `candidate`, and `screened` using the topic profile.
3. Produce digest, brief, gap report, paper cards, and evidence candidate outputs.
4. Keep all evidence in human-review mode. Do not mark `verified=true` automatically.
5. Produce a defensible related-work base for an AI/Big Data engineering thesis.

## Recommended first run

```bash
git clone https://github.com/SakJaeLim/paperops.git
cd paperops
git checkout topic/ax-ontology-governance

python -m venv .venv
# Windows
.venv\Scripts\activate
# macOS/Linux
# source .venv/bin/activate

pip install -r requirements.txt
python scripts/paperops.py init
python scripts/paperops.py collect --limit 30
python scripts/paperops.py score
python scripts/paperops.py screen --limit 120
python scripts/paperops.py digest --top 30
python scripts/paperops.py gap
python scripts/paperops.py brief
Comment on lines +42 to +43

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

P2 Badge Don't advertise stale gap and brief outputs

In this AX-specific runbook, these steps are presented as producing the gap report and thesis brief for the new governance topic, but scripts/paperops.py still hard-codes both cmd_gap and cmd_brief to the old "Ontology-enhanced GraphRAG ... 논문작성 지원" PaperOps/literature-writing thesis and meeting decisions about citation correctness/workflow efficiency, ignoring config/topic_profile.yaml. Users following the recommended sequence will therefore get reports framed around the previous topic rather than AX governance/security/decision automation; update those generators to read the profile or remove these steps from the AX runbook.

Useful? React with 👍 / 👎.

python scripts/paperops.py status
```

## If PDFs are needed

```bash
python scripts/paperops.py download-pdfs --limit 15
```

Optional GROBID parsing:

```bash
docker run -d -p 8070:8070 lfoppiano/grobid:0.8.0
python scripts/paperops.py parse-grobid --paper-id <paper_id> --apply
python scripts/paperops.py extract-evidence-candidates --paper-id <paper_id> --apply
python scripts/paperops.py review-evidence-candidates --paper-id <paper_id>
python scripts/paperops.py promote-evidence --paper-id <paper_id> --apply
python scripts/paperops.py guard-no-auto-verified --promoted-only
python scripts/paperops.py smoke-test
```

## Thesis framing to use while reviewing papers

### Core research question

How can an ontology-based operating model improve traceability, governance, security control, and decision automation in enterprise AX systems using generative AI?

### Sub-questions

1. What structural gap exists between conventional RAG/LLM applications and ontology/KG-based operational systems?
2. Which ontology components are required for AX operations: object, relation, policy, state, action, evidence, user role, risk, control, decision, and audit log?
3. How can SHACL/rule constraints and access-control metadata reduce unsafe or non-compliant LLM actions?
4. How should GraphRAG and agentic workflows be evaluated beyond answer accuracy, using traceability, reproducibility, governance compliance, and decision quality?
5. What reference architecture can integrate ontology/KG, RAG, LLM agents, governance controls, security guardrails, and human approval gates?

### Candidate contribution

A reference architecture and evaluation framework for ontology-based AX operating systems that combines:

- enterprise ontology / knowledge graph
- GraphRAG context construction
- policy and security constraints
- decision automation workflow
- human-in-the-loop approval
- audit and lineage tracking
- engineering evaluation metrics

## Review lens

For each paper, extract only the following:

1. Problem definition
2. Limitation of existing methods
3. Engineering contribution
4. Dataset/system setting
5. Evaluation metrics
6. Relevance to AX ontology governance
7. Whether it supports architecture, governance, security, or decision automation

## Expected thesis chapter skeleton

1. Introduction
- AX systems need controlled, traceable, and governable generative AI.
2. Related Work
- Ontology/KG, GraphRAG, AI governance, LLM security, decision automation, LLMOps.
3. Problem Definition
- Conventional LLM/RAG systems lack operational semantics, policy binding, and auditability.
4. Proposed Architecture
- Ontology-based AX operating model with KG, GraphRAG, agent workflow, policy/rule layer, security guardrails, and approval gates.
5. Implementation / Prototype
- Domain ontology, graph schema, retrieval pipeline, decision workflow, logging, and validation rules.
6. Evaluation
- Accuracy, groundedness, traceability, rule violation rate, decision reproducibility, latency, and human-review efficiency.
7. Discussion
- Enterprise applicability, limitations, governance risks, and future work.
8. Conclusion

## Immediate manual gates

After the first run, inspect:

- `reports/daily_digest/`
- `reports/survey_reports/`
- `data/metadata/papers.sqlite`
- `matrices/evidence_matrix.csv`
- `logs/ACTIVITY_LOG.md`

Do not promote evidence until the paper has been manually checked.