Neurosymbolic Language Reasoning as Satisfiability Modulo Theory

TL;DR

This paper introduces Logitext, a neurosymbolic language that combines natural language constraints with SMT solving to improve reasoning in natural language understanding tasks, extending neurosymbolic methods beyond formal domains.

Contribution

It presents Logitext, a novel framework that integrates LLMs with SMT solvers for joint textual and logical reasoning on natural documents.

Findings

Improves accuracy and coverage on content moderation tasks.

Extends neurosymbolic reasoning to partial logical structures in natural language.

First approach to treat LLM reasoning as an SMT theory.

Abstract

Natural language understanding requires interleaving textual and logical reasoning, yet large language models often fail to perform such reasoning reliably. Existing neurosymbolic systems combine LLMs with solvers but remain limited to fully formalizable tasks such as math or program synthesis, leaving natural documents with only partial logical structure unaddressed. We introduce Logitext, a neurosymbolic language that represents documents as natural language text constraints (NLTCs), making partial logical structure explicit. We develop an algorithm that integrates LLM-based constraint evaluation with satisfiability modulo theory (SMT) solving, enabling joint textual-logical reasoning. Experiments on a new content moderation benchmark, together with LegalBench and Super-Natural Instructions, show that Logitext improves both accuracy and coverage. This work is the first that treats LLM-based reasoning as an SMT theory, extending neurosymbolic methods beyond fully formalizable domains.