Decision Synthesis
Read-Only E1-E3 algorithmic composition + LLM refinement with Toulmin structurePrompts
Settings
DP from Causal Links (fallback)
strategies.py :: _llm_generate_from_causal_links()
Loading...
DP from Q&C Direct (last resort)
strategies.py :: _llm_generate_from_qc_direct()
Loading...
Decision Refinement
strategies.py :: _build_refinement_prompt()
Loading...
Board Choice Verification
board_choice_verifier.py :: verify_board_choices()
Loading...
Obligation Decision-Relevance (E1 fallback)
obligation_coverage_analyzer.py :: _llm_identify_decision_relevant()
Loading...
Algorithmic Stages (No LLM)
These run before LLM refinement
E1: Obligation Coverage
E2: Action-Option Mapping
E3: Decision Composition
Q&C Alignment Scoring
Decision Synthesis Parameters
Minimum Q&C alignment score for candidates (0.0-1.0)
Max candidates to refine via LLM
Minimum confidence for provision detection
Use only LLM for decision synthesis
LLM Parameters
Model for synthesis LLM calls
0.0=deterministic, 1.0=creative
Maximum response tokens
Phase Info
Service File
app/services/decision_point_synthesizer/strategies.py
Prompts
5 LLM prompts
Algorithmic Stages
4 stages (no LLM)
Test with Case
Go to Step 4 page
Select a case above first
Step 4 Pipeline
Provisions
Precedents
Questions
Conclusions
Transformation
Rich Analysis
Decision Synthesis
Narrative