3.2 KiB
3.2 KiB
ARC Database & Interface TODO
✅ Completed
ConceptARC Corpus
- Add ConceptARC corpus to repository
- Upload ConceptARC to database (160 puzzles, 16 categories)
- Categorize puzzles in DB (stored in
setcolumn) - Fix solution extraction for multiple test cases (array of solutions)
- Remove duplicate entries from arc_puzzles table (260 duplicates cleaned)
- Verify solution formats across all corpora
🔄 In Progress
Database
Fixes
- Corpora Mixup (V1 vs V2)
- Make id the index
Views & Queries
- Create database view with skills/concepts mapping
- Map ConceptARC categories to skill types
Data Quality
- Verify all V1/V2 puzzles have solutions
- Validate JSON integrity across all puzzles
- Document database schema
Schema Updates
- Add created_at timestamps to relevant tables
- Create N-to-N tags table structure
- Create tags table
- Create arc_puzzle_tags junction table
Retool Interface
Core Features
- Update puzzle renderer to handle ConceptARC multiple solutions
- Show all 3 test outputs for ConceptARC
- Remove header frame, consolidate to single-frame interface
- Add puzzle filtering by:
- Corpora (V1, V2, ConceptARC)
- Category/Set
- Difficulty
- Skills/Concepts
User Features
- Implement user input system
- Design input capture mechanism
- Store user solutions in database
- Create puzzle assignment system
- Add assignment toggle in interface
- Assign puzzles to specific users
- Track completion status
- Record time spent on puzzles
Tag System
- Design tag taxonomy
- Define tag types (concept, difficulty, skill, etc.)
- Implement tag database schema
- Create tags table
- Create puzzle_tags junction table
- Add tag management endpoints
- Build tag UI in Retool
- Tag creation/editing interface
- Bulk tagging functionality
- Tag-based search/filtering
- Test skill selection with many items (search function?)
- Populate initial tags
- Auto-tag ConceptARC by category
- Tag V1/V2 by known concepts
Administration
- Publish Retool app to production
📋 Backlog
Documentation
- Create user guide for Retool interface
- Document API/database schema
Testing
- Test puzzle rendering across all corpora
- Validate solution checking logic
- Test multi-user concurrent access
- Performance testing with full dataset
📝 Notes
Database Stats (Current)
- Total puzzles: 2,400
- V1: 1,120
- V2: 1,120
- ConceptARC: 160 (16 categories × 10 puzzles)
- All puzzles have solutions stored
- ConceptARC uses array format (3 solutions per puzzle)
Scripts Available
upload_conceptarc_to_db.py- Upload ConceptARC puzzlescheck_schema.py- Inspect database structurefind_duplicates.py- Find duplicate puzzle entriesremove_duplicates.py- Clean duplicate entriesfix_conceptarc_solutions.py- Fix ConceptARC solution formatverify_conceptarc_upload.py- Verify upload integrityverify_solution_format.py- Check solution formats