FAQ Generator
The FAQ Generator analyzes conversation history to identify frequently asked questions and automatically creates cached responses for them.
What It Does
The FAQ Generator:
- Queries your conversation history
- Finds questions asked multiple times
- Filters by success rate (good feedback)
- Generates high-quality cached responses
- Stores them for instant retrieval
Key Point
Your system has automatic caching that runs continuously. The FAQ Generator is an optional optimization tool for pre-populating cache or bulk processing. Your chatbot works perfectly without manually running this.
When to Use
Good Times to Run
- Initial setup: Pre-populate cache with common questions
- After content updates: Generate caches for new material
- Semester start: Prepare for anticipated questions
- Low cache hit rate: Improve coverage
Skip If
- Cache hit rate is already good
- System is performing well
- You've recently run it
Configuration Options
Limit (Top N Questions)
How many top questions to process, ordered by frequency.
| Value | Use Case |
|---|---|
| 10-25 | Quick test or targeted caching |
| 50 | Standard run (recommended) |
| 100+ | Comprehensive caching |
Recommendation: Start with 50 for initial runs.
Min Times Asked
Only cache questions asked at least this many times.
| Value | Effect |
|---|---|
| 1 | Cache everything (not recommended) |
| 2 | Cache repeat questions (default) |
| 5+ | Only very common questions |
Recommendation: Use 2-3 to ensure genuine demand.
Min Success Rate
Success rate = positive feedback / total feedback. Only caches questions with good user satisfaction.
| Value | Meaning |
|---|---|
| 0.3 | 30% positive - includes mixed questions |
| 0.5 | 50% positive - balanced (default) |
| 0.7 | 70% positive - high quality only |
Recommendation: Use 0.5 for balanced results.
How to Use
Step 1: Configure Settings
- Set Limit to your desired number (start with 50)
- Set Min Times Asked (default 2 is good)
- Set Min Success Rate (default 0.5 is good)
Step 2: Run Dry Run First
- Click Dry Run (Preview)
- Review the output panel
- See which questions would be cached
- Verify the numbers look reasonable
Step 3: Generate FAQs
- If dry run looks good, click Generate FAQs
- Watch the progress in the output panel
- Review results when complete
Understanding the Output
The output panel shows:
📊 Results:
Questions analyzed: 50
Successfully cached: 35
Skipped (low quality): 10
Failed: 5
📝 Details:
- Cached: "What is HFIM?" (asked 25 times)
- Skipped: "random question" (low success rate)
...
Result Categories
| Category | Meaning |
|---|---|
| Questions analyzed | Total questions reviewed |
| Successfully cached | New cache entries created |
| Skipped | Didn't meet quality threshold |
| Failed | Error during generation |
After Running
Verify Results
- Go to Cache Management
- Sort by "Created Date" descending
- Review newly created entries
- Edit any that need improvement
Monitor Performance
- Check Analytics over the next week
- Look for improved cache hit rate
- Note which generated entries are being used
Best Practices
Start Conservative
- Use smaller limits first (25-50)
- Use higher success rate threshold (0.5-0.7)
- Expand after reviewing results
Review Generated Content
- Generated answers may need refinement
- Check sources are accurate
- Add variations to improve matching
Don't Over-Generate
- Running repeatedly doesn't help
- Wait between runs to accumulate new data
- Focus on quality over quantity
Troubleshooting
No Questions Found
- Check your min times asked setting
- Try a larger time window
- Verify conversations exist in the database
Low Success Rate
- Your feedback data may be limited
- Lower the min success rate threshold
- Or wait for more user feedback
Many Skipped
- Quality threshold may be too high
- Lower min success rate if appropriate
- Check conversation quality
Generation Failed
- Check the error messages
- May indicate API issues
- Try again later
Related: