We wanted to test whether LLMs and decision models, specifically Jev and Luna, could solve a hard problem that we encounter regularly (grading identity resolution results). We found that while both models performed okay out of the box (88.8–97.3% accuracy), implementing a hybrid model that combines Jev + Opus achieves frontier accuracy at > 4x lower costs.
Identity resolution is the process of mapping data to corresponding real people. More precisely, say you have a set of data records .css-1n8ftex{background:var(--chakra-colors-brand-backgroundSecondary-light);color:var(--chakra-colors-brand-textSecondary-light);border-radius:var(--chakra-radii-tinier);font-family:var(--chakra-fonts-mono);white-space:pre-wrap;-webkit-padding-start:var(--chakra-space-1);padding-inline-start:var(--chakra-space-1);-webkit-padding-end:var(--chakra-space-1);padding-inline-end:var(--chakra-space-1);padding-top:var(--chakra-space-0-5);padding-bottom:var(--chakra-space-0-5);}.css-1n8ftex span{overflow-wrap:anywhere;}.css-1n8ftex:not(.mdx-heading .css-1n8ftex){font-size:var(--chakra-fontSizes-sm);}D. Determine a set of identities I and mapping m(D)->I such that each record d ∈ D maps to an identity i ∈ I , each i corresponds to exactly one real person in the real world, and no two i correspond to the same person in the real world.
