# GENERATE TARGETED SUMMARY PROCESS (Data-Centric, AI Proposes & User Confirms) ## Process Steps: 1. AI: Verify Schema access. Access Project State & Task Instance. 2. AI: Extract source text reference and summary parameters. 3. AI: Formulate yes/no questions to confirm parameters. 4. AI: Present questions. User confirms/corrects. If correction, loop back. 5. AI: Perform summarization based on confirmed parameters. Apply rigor. 6. AI: (Optional) Internal critique & self-correction. 7. AI: Prepare draft `summary_data_object`. 8. AI: Present proposed summary text and ask for acceptance (Yes/No). 9. User: Responds. If "No"/corrects, AI returns to step 5 to revise. 10. AI: If "Yes", finalize `summary_data_object` and store in `task_execution_instance.output_data`. Capture provenance if needed. Update task status. Update metadata timestamp. 11. AI: Formulate and present completion statement. Signal completion to calling process. ## Example Structured Questions (AI to formulate): * "Confirm source text is [Reference]? (Yes/No)" * "Confirm summary type is [Abstractive/Extractive/Hybrid]? (Yes/No)" * "Confirm length constraint is [Constraint]? (Yes/No)" * "Confirm focus is [Focus]? (Yes/No)" * **[After generation]:** "I have drafted the summary: '[Summary Text]'. Does this meet the specified parameters and your expectations? (Yes/No - If No, please provide feedback for revision)." * **[IF rejected]:** "Understood. Please provide specific feedback on how the summary should be revised." (Freeform expected) * **[IF completed]:** "Targeted summary generation complete. Output stored in project state for Task Execution ID [ID]." ---