Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Neuro-Symbolic Compliance: Integrating LLMs and SMT Solvers for Automated Financial Legal Analysis

Yung-Shen Hsia, Fang Yu, Jie-Hong Roland Jiang

TL;DR

The paper addresses the challenge of automated, verifiable financial regulatory compliance by proposing a neuro-symbolic framework that combines Large Language Models with Satisfiability Modulo Theories solvers. It introduces an end-to-end pipeline with Retrieval-Augmented Generation, SMT constraint generation (law and case facts), formal verification, and Weighted MaxSAT optimization within a modular multi-agent platform. Empirical results on 87 Taiwan FSC enforcement cases show high SMT-code generation accuracy ($\approx 86\%$), substantial speedups, and perfect compliance restoration in the evaluated scenarios, highlighting improved precision and auditability over LLM-only baselines. The work demonstrates a pathway toward scalable, verifiable regulatory automation with potential for cross-jurisdictional adaptation and human-in-the-loop oversight.

Abstract

Financial regulations are increasingly complex, hindering automated compliance-especially the maintenance of logical consistency with minimal human oversight. We introduce a Neuro-Symbolic Compliance Framework that integrates Large Language Models (LLMs) with Satisfiability Modulo Theories (SMT) solvers to enable formal verifiability and optimization-based compliance correction. The LLM interprets statutes and enforcement cases to generate SMT constraints, while the solver enforces consistency and computes the minimal factual modification required to restore legality when penalties arise. Unlike transparency-oriented methods, our approach emphasizes logic-driven optimization, delivering verifiable, legally consistent reasoning rather than post-hoc explanation. Evaluated on 87 enforcement cases from Taiwan's Financial Supervisory Commission (FSC), the system attains 86.2% correctness in SMT code generation, improves reasoning efficiency by over 100x, and consistently corrects violations-establishing a preliminary foundation for optimization-based compliance applications.

Neuro-Symbolic Compliance: Integrating LLMs and SMT Solvers for Automated Financial Legal Analysis

TL;DR

The paper addresses the challenge of automated, verifiable financial regulatory compliance by proposing a neuro-symbolic framework that combines Large Language Models with Satisfiability Modulo Theories solvers. It introduces an end-to-end pipeline with Retrieval-Augmented Generation, SMT constraint generation (law and case facts), formal verification, and Weighted MaxSAT optimization within a modular multi-agent platform. Empirical results on 87 Taiwan FSC enforcement cases show high SMT-code generation accuracy (), substantial speedups, and perfect compliance restoration in the evaluated scenarios, highlighting improved precision and auditability over LLM-only baselines. The work demonstrates a pathway toward scalable, verifiable regulatory automation with potential for cross-jurisdictional adaptation and human-in-the-loop oversight.

Abstract

Financial regulations are increasingly complex, hindering automated compliance-especially the maintenance of logical consistency with minimal human oversight. We introduce a Neuro-Symbolic Compliance Framework that integrates Large Language Models (LLMs) with Satisfiability Modulo Theories (SMT) solvers to enable formal verifiability and optimization-based compliance correction. The LLM interprets statutes and enforcement cases to generate SMT constraints, while the solver enforces consistency and computes the minimal factual modification required to restore legality when penalties arise. Unlike transparency-oriented methods, our approach emphasizes logic-driven optimization, delivering verifiable, legally consistent reasoning rather than post-hoc explanation. Evaluated on 87 enforcement cases from Taiwan's Financial Supervisory Commission (FSC), the system attains 86.2% correctness in SMT code generation, improves reasoning efficiency by over 100x, and consistently corrects violations-establishing a preliminary foundation for optimization-based compliance applications.
Paper Structure (24 sections, 8 equations, 3 figures, 6 tables, 3 algorithms)

This paper contains 24 sections, 8 equations, 3 figures, 6 tables, 3 algorithms.

Table of Contents

  1. Introduction
  2. Related Work
  3. Methodology
  4. SMT Constraint Generation
  5. Case Crawling and Preliminary Article Extraction
  6. Legal Article Retrieval and Parsing
  7. Supplementary Article Retrieval (RAG-Based Enrichment)
  8. SMT Constraint Generation
  9. Build Law Constraints
  10. Build Compliance Fact Constraints
  11. SMT Constraint Consistency Check and Optimization
  12. Law Constraint Consistency Check.
  13. Case Illegality Check.
  14. Optimization for Minimal Compliance Solution
  15. Multi-Agent Compliance Platform
  16. ...and 9 more sections

Figures (3)

  • Figure 1: SMT Constraint Generation Pipeline
  • Figure 2: System architecture of the Finance Compliance Analysis Agent.
  • Figure 3: Relationship between variables and constraints (a), and the distribution of hard, soft, and total constraints (b).