Theory-Level Autoformalization: From Isolated Statements to Unified Formal Knowledge Bases
Quick Answer
The paper advocates for theory-level autoformalization, emphasizing the need to formalize entire theories with interdependencies rather than isolated statements.
Quick Take
This approach aims to create structured libraries of formal knowledge, addressing challenges and proposing future research directions.
Key Points
- Theory-level autoformalization focuses on formalizing complete theories.
- It aims to create structured libraries of axioms, definitions, and lemmas.
- The paper identifies open challenges in the autoformalization process.
- Three promising research paths for future exploration are proposed.
- The significance of this shift in formalization efforts is discussed.
Paper Resources
📖 Reader Mode
~2 min readAbstract:Autoformalization translates informal natural language into formal, machine-verifiable languages. While most work focuses on individual statements, real formalization efforts are inherently theory-level: they require an entire web of axioms, definitions, and lemmas before target theorems can even be stated. In this position paper, we argue for theory-level autoformalization: formalizing complete theories, including all their inter-dependencies, as structured libraries. We examine the significance of this shift, address alternative views, identify open challenges, and propose three promising paths forward. Our survey of autoformalization is available at this https URL.
| Comments: | ICML 2026 Spotlight |
| Subjects: | Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG); Programming Languages (cs.PL) |
| MSC classes: | 68 |
| ACM classes: | F.4; I.2 |
| Cite as: | arXiv:2607.13292 [cs.AI] |
| (or arXiv:2607.13292v1 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2607.13292 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Marcus J. Min [view email]
[v1]
Tue, 14 Jul 2026 21:58:52 UTC (8,020 KB)
— Originally published at arxiv.org
Want this in your inbox every morning?
Daily brief at your local 8am — bilingual EN/中文, free.
More from arXiv cs.AI
See more →HOBA: Hierarchical On-Policy Bidding Agents for Adaptive Online Advertising
HOBA (Hierarchical On-policy Bidding Agents) is a novel hierarchical reinforcement learning framework that enhances online advertising bidding systems by improving adaptability and reducing hyperparameter tuning costs. It utilizes a for hyperparameter inference, a SARSA agent for expert model selection, and a dynamic expert pool for bid execution, achieving a +3.6% increase in target cost during large-scale deployment and outperforming state-of-the-art baselines on AuctionNet.