FAQ Question Pattern Analysis Workflow

Systematic process for generating and validating FAQ content from course materials

flowchart TD Start(("Course Materials
Assembled")):::startNode Extract["Extract Concept List
from Learning Graph"]:::autoNode Analyze["Analyze Concept
Dependencies"]:::autoNode Search["Search for
Question Patterns"]:::autoNode Generate["Generate Candidate
Questions (5-10 per concept)"]:::autoNode QualityCheck{"Quality
Threshold Met?"}:::decisionNode Flag["Flag for
Human Review"]:::reviewNode AddDB["Add to FAQ
Database"]:::autoNode Review["Educator Review
& Edit"]:::reviewNode GenAnswers["Generate Answers
from Corpus"]:::autoNode Validate["Validate Answer
Completeness"]:::validNode Complete{"Answer
Complete?"}:::decisionNode Revise["Revise Answer"]:::reviewNode Approve["Approve FAQ Entry"]:::validNode Update["Update FAQ Index"]:::autoNode End(("FAQ Published")):::endNode Start --> Extract Extract --> Analyze Analyze --> Search Search --> Generate Generate --> QualityCheck QualityCheck -->|No| Flag QualityCheck -->|Yes| AddDB Flag --> Review Review --> AddDB AddDB --> GenAnswers GenAnswers --> Validate Validate --> Complete Complete -->|No| Revise Revise --> GenAnswers Complete -->|Yes| Approve Approve --> Update Update --> End classDef startNode fill:#667eea,stroke:#333,stroke-width:3px,color:#fff,font-size:16px classDef autoNode fill:#4facfe,stroke:#333,stroke-width:2px,color:#fff,font-size:16px classDef decisionNode fill:#ffd166,stroke:#333,stroke-width:2px,color:#333,font-size:16px classDef reviewNode fill:#f77f00,stroke:#333,stroke-width:2px,color:#fff,font-size:16px classDef validNode fill:#43aa8b,stroke:#333,stroke-width:2px,color:#fff,font-size:16px classDef endNode fill:#9b59b6,stroke:#333,stroke-width:3px,color:#fff,font-size:16px linkStyle default stroke:#999,stroke-width:2px,font-size:14px

Workflow Overview

This systematic workflow uses both automated analysis and human review to generate high-quality FAQ content from course materials including learning graphs, glossaries, and chapter content.

Workflow Phases

Phase 1: Automated Analysis

  1. Extract Concept List - Parse learning graph to enumerate all concepts
  2. Analyze Dependencies - Identify complex concepts (high in-degree) and foundational concepts (zero dependencies)
  3. Search Patterns - Scan corpus for existing questions and common interrogative structures
  4. Generate Questions - Use Claude API to create 5-10 questions per concept across categories:
    • Definitional: "What is..."
    • Procedural: "How do I..."
    • Troubleshooting: "Why doesn't..."
    • Comparative: "What's the difference between..."

Phase 2: Quality Check

Questions must meet criteria:

Passing rate: Typically 60-70% of auto-generated questions pass quality check

Phase 3: Human Review

Subject matter expert reviews flagged questions and:

Phase 4: Answer Generation & Validation

  1. Generate Answers - Claude retrieves relevant passages from course content and synthesizes comprehensive answers with chapter citations
  2. Validate Completeness - Check that answers directly address questions, stay within course scope, and match reading level
  3. Revise if Needed - Educator supplements or rewrites answers; may identify content gaps requiring new chapter sections
  4. Approve Entry - Finalized FAQ question-answer pair added to /docs/faq.md

Phase 5: Publication

Update FAQ Index - Add search keywords, concept tags, and navigation links; integrate into MkDocs site

Analytics Tracking - Monitor which questions receive most views to identify student needs

Quality Metrics

Target: 100-200 FAQ entries for a complete course

Coverage: Every concept should have at least 1 FAQ question

Human intervention: 30-40% of questions require educator review

Answer completeness: 95%+ of answers should be self-contained (no external resources needed)

Integration Benefits