Add mock Confluence pages for the 'cquisition-and-Credit-Risk' space, designed specifically to demonstrate the Fraud Model Explainability Assistant with Confluence integration.

docs/MOCK_CONFLUENCE_PAGES_README.md

@@ -0,0 +1,324 @@
+# Mock Confluence Pages for Demo
+
+This directory contains three comprehensive mock Confluence pages designed for demonstrating the Fraud Model Explainability Assistant with Confluence integration.
+
+---
+
+## Files Created
+
+### 1. fraud-model-governance.md
+**Purpose**: Model documentation, SR 11-7 compliance, validation reports
+
+**Key Sections**:
+- Model Inventory
+- XGBoost v3.2 Model Validation Report (detailed)
+- Performance metrics and backtesting
+- Fair lending analysis
+- Champion/Challenger analysis
+- SR 11-7 compliance documentation
+- Approval records
+
+**Example Queries This Answers**:
+- "Find the model validation report for XGBoost v3.2"
+- "What documentation do we have for model v3.2?"
+- "What are the performance metrics for our fraud model?"
+- "Show me the champion/challenger analysis"
+
+**Content Highlights**:
+- Comprehensive validation report with AUC, KS, precision, recall metrics
+- Fair lending testing results by protected class
+- Backtesting results showing model performance post-deployment
+- Phone type feature analysis (relevant to compliance queries)
+- SR 11-7 compliance checklist
+
+---
+
+### 2. compliance-policies.md
+**Purpose**: Fair lending policy, ECOA compliance, adverse action guidelines
+
+**Key Sections**:
+- Fair Lending Policy
+- ECOA Compliance Procedures
+- Adverse Action Guidelines with fraud-specific procedures
+- Synthetic ID Detection and Fair Lending
+- **Phone Type Features Policy** (critical for demo queries)
+- Model Feature Fair Lending Requirements
+- Regulatory Examination Preparation
+
+**Example Queries This Answers**:
+- "What does our fair lending policy say about synthetic ID detection?"
+- "What does our fair lending policy say about phone type features?"
+- "Check fair lending compliance for APP-55555 and cite relevant policies"
+- "How do we document this for the OCC exam?"
+- "What are our adverse action procedures for suspected fraud?"
+
+**Content Highlights**:
+- Detailed phone type features policy with fair lending analysis
+- Adverse impact ratios for phone type features by protected class
+- Synthetic ID detection fair lending requirements
+- Adverse action reason codes for fraud cases
+- Sample examiner Q&A responses
+
+---
+
+### 3. fraud-investigation-playbooks.md
+**Purpose**: Investigation procedures, fraud patterns, escalation procedures
+
+**Key Sections**:
+- Investigation Framework
+- **Synthetic ID Fraud Investigation** (with detailed checklist)
+- High-Risk Application Investigation
+- Identity Network Analysis
+- Escalation Procedures
+- Case Studies (3 detailed examples)
+- Fraud Patterns and Red Flags
+
+**Example Queries This Answers**:
+- "What are the procedures for escalating high-risk applications?"
+- "What should I check for synthetic ID fraud?"
+- "How do I investigate a high-risk application?"
+- "What are the red flags for bust-out fraud?"
+- "Show me examples of fraud investigation cases"
+
+**Content Highlights**:
+- Complete synthetic ID investigation checklist (6 steps)
+- Real investigation example with full documentation
+- Network fraud analysis procedures
+- Escalation path diagram
+- 3 case studies with lessons learned
+- Comprehensive red flag categories with severity ratings
+
+---
+
+## Using These Files for Demo
+
+### Option 1: Direct Use as Markdown
+The files are already in Markdown format and can be used directly with the confluence-ingestor if it supports Markdown input.
+
+### Option 2: Convert to Confluence Format
+If you need actual Confluence pages for the demo:
+
+1. **Manual Upload**:
+   - Create a Confluence space called "Acquisition-and-Credit-Risk"
+   - Create three pages with the content from these files
+   - Confluence will convert the Markdown automatically when you paste it
+
+2. **Automated Import**:
+   - Use Confluence's bulk import feature
+   - Convert Markdown to Confluence wiki format using pandoc:
+   ```bash
+   pandoc fraud-model-governance.md -f markdown -t confluence -o fraud-model-governance.confluence
+   ```
+
+3. **Mock Ingestion**:
+   - Use these files as mock data for testing the RAG pipeline
+   - Store them in a local directory and ingest directly if your system supports file-based ingestion
+
+### Option 3: Use as Mock Vector Database Content
+If you want to simulate Confluence without actually setting it up:
+
+1. Ingest these three Markdown files directly into ChromaDB
+2. Tag them with mock Confluence metadata:
+   - Space key: "Acquisition-and-Credit-Risk"
+   - Page IDs: fraud-model-governance (001), compliance-policies (002), fraud-investigation-playbooks (003)
+   - URLs: Mock Confluence URLs
+
+---
+
+## Demo Query Examples
+
+These mock pages are specifically designed to answer the example queries in the Gradio interface:
+
+### Query 1: Model Governance
+**User**: "Find the model validation report for XGBoost v3.2"
+
+**Expected Result**: Should retrieve fraud-model-governance.md and cite:
+- Model validation report section
+- Performance metrics (AUC 0.870, KS 0.542)
+- Fair lending testing results
+- Approval signatures and dates
+
+---
+
+### Query 2: Fair Lending - Synthetic ID
+**User**: "What does our fair lending policy say about synthetic ID detection?"
+
+**Expected Result**: Should retrieve compliance-policies.md and cite:
+- Synthetic ID Detection and Fair Lending section
+- Approved detection features (SSN validity, credit file characteristics)
+- Fair lending testing requirements
+- Prohibited methods
+- Quarterly disparate impact testing procedures
+
+---
+
+### Query 3: Fair Lending - Phone Type Features
+**User**: "What does our fair lending policy say about phone type features?"
+
+**Expected Result**: Should retrieve compliance-policies.md and cite:
+- Phone Type Features Policy section
+- Fair lending analysis table showing adverse impact ratios
+- Policy statement (must pass 80% rule, quarterly testing)
+- Approved phone type features (phone type flag, tenure, velocity)
+- Monitoring and governance requirements
+
+---
+
+### Query 4: Investigation Procedures
+**User**: "What are the procedures for escalating high-risk applications?"
+
+**Expected Result**: Should retrieve fraud-investigation-playbooks.md and cite:
+- Escalation Procedures section
+- When to escalate (fraud score > 700, network fraud, etc.)
+- Escalation path (Level 1 → Level 2 → Level 3 → Level 4)
+- Escalation process (4 steps)
+- Escalation contacts table with response times
+
+---
+
+### Query 5: Synthetic ID Investigation
+**User**: "What should I check for synthetic ID fraud?"
+
+**Expected Result**: Should retrieve fraud-investigation-playbooks.md and cite:
+- Synthetic ID Fraud Investigation section
+- Synthetic ID red flags (primary and secondary indicators)
+- 6-step investigation checklist
+- SSN validation procedures
+- Example investigation with full documentation
+
+---
+
+### Query 6: Compliance Check with Policy Citation
+**User**: "Check fair lending compliance for APP-55555 and cite relevant company policies"
+
+**Expected Result**: Should retrieve both compliance-policies.md AND fraud-model-governance.md:
+- Fair lending policy statement
+- ECOA compliance procedures
+- Model feature fair lending requirements
+- XGBoost v3.2 fair lending testing results
+- Specific adverse impact ratios for protected classes
+- Phone type features analysis
+
+---
+
+## Content Design Principles
+
+These mock pages were designed with the following principles:
+
+### 1. Realism
+- Corporate Confluence page style and formatting
+- Realistic policy language and regulatory references
+- Actual financial services terminology and frameworks
+- Proper compliance and governance structures
+
+### 2. Searchability
+- Rich keyword coverage for semantic search
+- Multiple ways to refer to the same concept
+- Specific examples that match Gradio query examples
+- Cross-references between pages
+
+### 3. Demo-Optimized
+- Content specifically answers the 9 example queries in the Gradio interface
+- Includes specific examples mentioned (XGBoost v3.2, phone type features)
+- Provides enough detail for impressive demo responses
+- Includes tables, metrics, and structured data
+
+### 4. Educational Value
+- Real fraud patterns and investigation techniques
+- Actual regulatory frameworks (SR 11-7, ECOA, Regulation B)
+- Genuine fair lending considerations
+- Professional best practices
+
+---
+
+## Statistics
+
+**Total Content**:
+- 3 comprehensive Confluence pages
+- ~23,000 words total
+- ~150KB of text content
+- 50+ sections across all pages
+- 15+ tables with structured data
+- 3 detailed case studies
+- 20+ code examples and templates
+
+**Coverage**:
+- Model governance: ✅ Complete
+- Fair lending compliance: ✅ Complete
+- Fraud investigation procedures: ✅ Complete
+- Synthetic ID detection: ✅ Extensive coverage
+- Phone type features: ✅ Detailed policy
+- Escalation procedures: ✅ Complete workflows
+- Regulatory examination prep: ✅ Sample Q&A
+
+---
+
+## Next Steps
+
+### For Demo Setup:
+
+1. **Copy files to appropriate location**:
+   ```bash
+   # If using local file ingestion
+   mkdir -p demo_confluence_pages
+   cp fraud-*.md compliance-*.md demo_confluence_pages/
+   ```
+
+2. **Ingest into RAG pipeline**:
+   ```python
+   from confluence_ingestor import ConfluenceRAG
+
+   # Option A: If using actual Confluence (after uploading)
+   rag = ConfluenceRAG.from_env()
+   rag.ingest_space("Acquisition-and-Credit-Risk", max_pages=10)
+
+   # Option B: If using local files (modify as needed)
+   # Implement file-based ingestion using your RAG pipeline
+   ```
+
+3. **Test with example queries**:
+   - Run each of the 9 Gradio example queries
+   - Verify relevant sections are retrieved
+   - Check citation accuracy (page titles, URLs)
+
+4. **Fine-tune retrieval**:
+   - Adjust `k` parameter (number of results)
+   - Test space filtering
+   - Verify semantic search quality
+
+---
+
+## Maintenance
+
+To update the mock pages for future demos:
+
+1. **Add new content**: Follow existing structure and formatting
+2. **Update dates**: Change "Last Updated" dates to current
+3. **Add new examples**: Create additional queries in Gradio that these pages can answer
+4. **Expand case studies**: Add more fraud investigation examples
+5. **Update regulations**: Keep regulatory references current
+
+---
+
+## License and Usage
+
+**Usage**: These files are mock examples for demonstration purposes only.
+
+**Not for Production**: This is fictitious content designed for demos and testing. Do not use as actual compliance or governance documentation.
+
+**Customization**: Feel free to modify content to match your demo needs, add additional sections, or create additional mock pages.
+
+---
+
+## Questions?
+
+For questions about these mock Confluence pages or how to use them in your demo:
+- See CONFLUENCE_SETUP_GUIDE.md for integration instructions
+- See QUICK_START.md for 10-minute setup
+- See IMPLEMENTATION_SUMMARY.md for complete integration overview
+
+---
+
+**Created**: January 2026
+**Purpose**: Fraud Model Explainability Assistant Demo
+**Status**: Ready for use ✅