# GENERATE ACADEMIC PAPER PROCESS (Data-Centric, Structured Interaction, Peer Review Rigor & Provenance)
## Process Steps:
1. Access Project State & Task Instance.
2. Formulate understanding questions (Paper Section, Outline, Inputs, Standards, **Rigor & Provenance Requirements**).
3. Structure AI output turn (project ID + task execution ID + status + summary + questions).
4. Present structured AI output turn to user.
5. Receive user responses (handle freeform by returning to step 2).
6. Verify inputs and readiness; return to step 2 if not ready.
7. Perform Generation/Revision (**Absolute Rigor Focus: Fact-Checking, Citation, ZERO Hallucination**). **Actively Capture Provenance Data.**
8. Identify issues/clarifications (**Flag Unverified Claims/Conflicts**).
9. Formulate clarification questions (may require non-yes/no).
10. Structure AI output turn (project ID + task execution ID + status + issue summary + questions).
11. Present structured AI output turn to user.
12. Receive user responses (handle freeform by returning to step 8).
13. **Generate Collaboration Disclosure Text**, store in state.
14. Finalize Output Data (paper content reference + metadata), store in state.
15. **Finalize Provenance Data**, store in state.
16. Update Task Execution Instance status, log issues/insights, update feedback status in state.
17. Formulate completion questions (**Confirm Rigor Adherence, Confirm Disclosure Text, Confirm Provenance Prepared**).
18. Structure AI output turn (project ID + task execution ID + final status + outcome summary + questions).
19. Present structured AI output turn to user.
20. Receive user responses.
21. Update Project State based on final responses.
22. Indicate readiness for next process step (implicitly, by completing this process).
## Example Structured Questions (AI to formulate based on task details, inputs, and process progress):
* "Confirm understanding: Draft the Introduction section of the paper? (Yes/No)"
* **Confirm understanding: This task requires absolute adherence to fact-checking, strict citation, zero hallucination/fabrication, AND preparation of provenance data? (Yes/No)**
* "Clarification needed: Source '[Source Reference]' states X, but outline requires Y. Which should I follow? (Follow Source/Follow Outline/Other - please specify)"
* **Clarification needed: I could not definitively verify claim Z against provided sources. Should I include it with a caveat, or omit it? (Include with caveat/Omit/Other - please specify)**
* "Drafting of [Section Name] complete. Ready for user review? (Yes/No)"
* **Does the generated content adhere to the required rigor standards (fact-checking, citation, no hallucination) to the best of my ability? (Yes/No - AI self-assessment)**
* **Is the generated Collaboration Disclosure text acceptable? (Yes/No - If No, please provide feedback/revisions)**
* **Has the provenance data for this section/task been prepared? (Yes/No - AI self-assessment)**
* "Task [Task ID] output is ready. Initiate review process using [[archive/templates/projects/support/REVIEW-Feedback]]? (Yes/No)"
---
**Project State Data Schema Update (Implicit in instructions):**
The `DEFINE-ProjectStateSchema.md` would need to include the `provenance_data` field within the `academic_paper_data` structure, and potentially define its internal structure (e.g., a list of objects linking content segments to sources/inputs/process steps).
```yaml
# Structure for academic_paper_data within task_execution_instance.output_data
academic_paper_data:
section_title: string # e.g., "Introduction", "Abstract", "Full Paper"
content_reference: string # Reference to where the actual text is stored (user's file system)
word_count: integer
citation_count: integer
collaboration_disclosure_text: string # The generated disclosure statement
provenance_data: object # Structured data detailing the generation process for this content segment
# Example structure (needs formal definition in schema):
# content_segments: list of objects { segment_id: string, text_snippet: string (optional), generated_from: list of references (source_id, analysis_id, task_id, user_input_ref), process_steps: string (summary of AI's internal steps) }
# Add other relevant metadata like adherence scores, review status summary