ad

agentgraph/methods/production/openai_structured_extractor.py

@@ -105,6 +105,12 @@ NAMING STYLE & CANONICALIZATION:
 - Domain templates (use when content clearly matches):
   * Location-based discovery: Tasks → Geographic Proximity Analysis → Restaurant Data Collection → Operating Hours Validation; Output → Restaurant Recommendations; System → Location-Based Restaurant Discovery System
   * Wildlife data analysis: Tasks → Government Dataset Processing → Population Statistics Calculation → Ecological Data Validation; Human → Wildlife Researcher; System → Wildlife Data Analysis and Ecological Monitoring System
+  * Enterprise QA / Document QA (tags/keywords: aif, inference, classification, decomposition, rephrasing, retrieval, answer synthesis, business response):
+    - Tasks (3–5 max, choose those evidenced in the trace):
+      Query Classification → Subquestion Decomposition and Rephrasing → Context Retrieval and Assembly → Domain Question Answering → (optional) Review/Validation
+    - Agents (examples): Query Classifier, Question Rephrasing Specialist, Oxford Economics Assistant
+    - Input/Output: User Query → Answer Response (or domain-specific final artifact)
+    - Tool: Computer Terminal (model/API execution)
 
 == FEW-SHOT EXAMPLE (COMPLEX DISCOVERY WORKFLOW) ==
 
@@ -177,6 +183,7 @@ ANALYSIS STEPS:
      * Contains "cost", "savings", "ticket", "verification" → VERIFICATION (3 specialized tasks)
      * Contains "location", "restaurant", "proximity", "search" → DISCOVERY (3 sequential tasks)  
      * Contains "probability", "game theory", "chemistry" → INTERDISCIPLINARY (3 domain tasks)
+     * Contains "classification", "decompose", "retrieval", "answer", "business response", "Oxford Economics" → ENTERPRISE_QA/DOC_QA (3–5 tasks from the template)
      * Simple single-agent scenarios → SIMPLE (1 agent, 1 task)
    
    - GENERATE tasks accordingly (Independent Task Allocation):
@@ -214,6 +221,7 @@ ANALYSIS STEPS:
      * Output→Human (DELIVERS_TO): 1 relation only
      * Agent→Tool (USES): Include exactly one when a Tool exists
      * Tool→Task (REQUIRED_BY): Include one only when the task explicitly depends on a Tool (execution/platform dependency). Otherwise omit.
+     * Task→Task (SUBTASK_OF): Use sparingly when the trace explicitly states that one task is a component inside another (e.g., Decomposition inside Question Answering). Do not replace NEXT; both may coexist when justified.
    
    - FORBIDDEN: ASSIGNED_TO, INTERVENES
    - TARGET: Maximum 9–10 total relations for 3-agent workflows
@@ -229,6 +237,9 @@ NAMING & CANONICALIZATION:
 - Title Case + spaces only (no underscores in names)
 - Prefer canonical domain templates (see system rules) when content/tags match
 - Prefer concise outputs like "Restaurant Recommendations" for discovery workflows
+ - For Enterprise QA/Document QA, prefer the following canonical names when supported by evidence:
+   * Query Classification, Subquestion Decomposition and Rephrasing, Context Retrieval and Assembly, Domain Question Answering
+   * Input: User Query; Output: Answer Response
 
 FORMATTING:
 - IDs: agent_001, task_001, tool_001, etc.