Status:

• Sunny/Filipe showcased on Feb 12 in a local dev environment
• We will repeat the same Showcase with 3 changes: (1) In a deployed environment; (2) 50-row CIEL dataset prepared by Filipe instead of 10-row handpicked dataset; (3) Run by Andy instead of Sunny
• Dataset for this upcoming showcase: OpenConceptLab/ocl_issues#2340
• Sunny to walk Andy through how to run this showcase (this is mostly about how to configure the custom match algorithms)
• This should be just a straight run-through and video recorded - aiming for 5-10min

Goal:

• Demo of the icd11 evaluation end-to-end workflow with a test dataset by Feb 12
• Audience: Local team, but intending to expand to WHO/HL7/whoever
• Demo Driver: Andy main (previously Filipe), Sunny support

Steps for End-to-end workflow demo:

1. Start new Project in OCL Mapper
2. Load Input Dataset: test dataset formatted as CSV
3. Configure project:

• Target repo: WHO / ICD-11-WHO (Not defined in OCL)
• Algorithms:
  • ocl-ciel-bridge
  • icd11 llm-as-terminologist
  • icd11 auto-code
• Reranker: BAAI/bge-reranker-v2-m3 (this is the current default re-ranker model for OCL)
• Field Mapping: Only match on the name
• AI Assistant:
  • Exclude mapping fields
  • Model: anthropic/claude-haiku-4.5

4. Auto-match: 3x algorithms, reranker, AI assistant
5. Save project
6. Download project as CSV