Context
The fallacy taxonomy contains nodes that are not individual cards (family headers, order groupings, etc.). These nodes need content enrichment (descriptions, examples, cross-references) for the website and ontology.
A substantial script was developed ~2 years ago using GPT-4 API to semi-automatically enrich taxonomy nodes. This script needs to be adapted to work with:
- Updated OpenAI/Anthropic APIs
- Modern models (Claude 4.x, GPT-4o/4.5)
- Current CSV structure
Scope
1. Identify Non-Card Nodes
- Parse the Fallacies taxonomy CSV
- Identify nodes that are families/orders (not leaf cards)
- Catalog which fields are empty/incomplete
2. Locate and Adapt the Enrichment Script
- Find the original GPT-4 script in the repo or related repos
- Update API calls to current SDK versions
- Consider using Claude API for better quality/cost ratio
- Add batch processing with rate limiting
3. Enrichment Targets
- Node descriptions (FR + translations)
- Real-world examples
- Cross-references between related fallacies
- Educational content for the website
Dependencies
🤖 Generated with Claude Code
Context
The fallacy taxonomy contains nodes that are not individual cards (family headers, order groupings, etc.). These nodes need content enrichment (descriptions, examples, cross-references) for the website and ontology.
A substantial script was developed ~2 years ago using GPT-4 API to semi-automatically enrich taxonomy nodes. This script needs to be adapted to work with:
Scope
1. Identify Non-Card Nodes
2. Locate and Adapt the Enrichment Script
3. Enrichment Targets
Dependencies
🤖 Generated with Claude Code