Explains probabilistic record linkage
Fellegi-Sunter: agreement weights per field, an upper and a lower threshold, and a clerical-review region between them; discriminating fields separate people who share common values.
Can AI do this?
AI does this well Most of the tested AI models master this. 3 of 4 models in the current benchmark master it.
| Model | Tier | Probability of mastery | Status |
|---|---|---|---|
| gemini-4-argon | Flagship | 100% | Mastered |
| gemini-3.5-flash-lite | Low cost | 95% | Not shown |
| gpt-6-luna | Low cost | 100% | Mastered |
| claude-haiku-5-5 | Low cost | 100% | Mastered |
Mastered means a probability of at least 90% that the model has this knobit, estimated from its graded answers. “Not shown” means the answers do not settle it either way. How the models are tested.
What it looks like in practice
Expected result: Fellegi-Sunter: agreement weights per field, an upper and a lower threshold, and a clerical-review region between them; discriminating fields separate people who share common values.
How it is assessed: Answers a written question on the record linkage correctly
Related knobits
Builds on
Needed for
- R-302 Recognises a fuzzy rule merging household members
- R-303 Recognises auto-merge with no clerical-review band
- R-304 Recognises match results depending on rule order
- R-308 Recognises no discriminator for same-name relatives
- R-331 Requires a discriminator before merging same-name relatives
- R-332 Introduces a clerical-review band below auto-merge
- R-333 Decides record pairs without false merges
All 16 knobits of Define the matching rules.
Source: Practice the stage 3 reference and review checklist rest on