Policy-as-logic for robust reasoning over rules
On the map: LLM reasoning → Reasoning benchmarks and evaluation
Abstract (arXiv)
In many practical applications of generative AI systems, from tax rules to airline baggage allowance, responses to natural language queries must respect written policies or rules. We present a hybrid symbolic approach that expresses policies in formal logic and at inference time exploits the representation power of language models for fact extraction to ground predicates, and an answer set solver for reasoning such that responses are interpretable, auditable, and as we show, accurate and robust under input perturbations. Specifically, we show this separation of extraction and reasoning steps outperforms policy-as-prompt and policy-as-code methods in most cases with ~10x reduction in token usage. The results point to the value of structured reasoning and symbolic solvers in conjunction with generative models to make robust decisions involving objective criteria.
Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in Reasoning benchmarks and evaluation, one of 17 directions in LLM reasoning. Builds on 8 of its 20 references that are in the map. First posted on arXiv 12 Aug 2026.
Builds on
8 of 20 references.
Closest ancestors · 3
Same direction, newest first
- 2026CL-bench: A Benchmark for Context Learning3 Feb
- 2025Logical Reasoning in Large Language Models: A Survey13 Feb
- 2023From Word Models to World Models: Translating from Natural Language to the Probabilistic Language of Thought22 Jun
Other references · 5
- 2024Justice or Prejudice? Quantifying Biases in LLM-as-a-Judgecited by 368
- 2025PolyGuard: A Multilingual Safety Moderation Tool for 17 Languagescited by 59
- 2014Clingo = ASP + Control: Preliminary Reportcited by 56
- 2024Granite Guardiancited by 25
- 2026DeonticBench: A Benchmark for Reasoning over Rulescited by 5
Built on by
Not cited yet.
Similar, not cited
Close in wording, no citation either way
- 2026PL-Guard: Probabilistic Logic Reasoning for LLM Guardrails
- 2025Policy-Aware Generative AI for Safe, Auditable Data Access Governance
- 2025A Neurosymbolic Approach to Natural Language Formalization and Verification
- 2025From Language to Logic: A Bi-Level Framework for Structured Reasoning
- 2026Pushing the Boundaries of Natural Reasoning: Interleaved Bonus from Formal-Logic Verification
- 2026ORACLE: Optimizing Reasoning Abilities of Large Language Models via Constraint-Led Synthetic Data Elicitation
- 2025A Novel Architecture for Symbolic Reasoning with Decision Trees and LLM Agents
- 2024Scaling Synthetic Logical Reasoning Datasets with Context-Sensitive Declarative Grammars