Lean Finder: Semantic Search for Mathlib That Understands User Intents
Jialin Lu, Kye Emond, Kaiyu Yang, Swarat Chaudhuri, Weiran Sun, Wuyang Chen · 2025
Read Lean Finder: Semantic Search for Mathlib That Understands User Intents by Jialin Lu, Kye Emond, Kaiyu Yang, Swarat Chaudhuri, Weiran Sun, Wuyang Chen (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2510-15940.txt.
Description
Jialin Lu and five co-authors present Lean Finder, a semantic search engine for the Lean proof system and its mathlib library. Formal proving stalls when relevant theorems are hard to find, and Lean's steep learning curve slows newcomers further. Older search tools lean on informal translations of statements, which often mismatch how users actually ask.
The team starts from real user intent: they analyse and cluster public Lean discussions, then fine-tune text embeddings on synthetic queries that mimic those intents. Diverse feedback signals align the engine with mathematicians' preferences from several angles. On real queries, informal statements, and proof states, Lean Finder improves over earlier engines and a leading general model by a large relative margin. It also plugs into language-model provers, joining retrieval with formal reasoning.
Catalog record
| CREATOR | Jialin Lu, Kye Emond, Kaiyu Yang, Swarat Chaudhuri, Weiran Sun, Wuyang Chen |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 98 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2510-15940