Integrating Logic Programming with Large Language Models: Opportunities and Challenges
Integrating Logic Programming with Large Language Models: Opportunities and Challenges

Author(s): Miglena Stoyanova
Subject(s): Economy, Supranational / Global Economy, Business Economy / Management, ICT Information and Communications Technologies, Transport / Logistics
Published by: Икономически университет - Варна
Keywords: logic programming; large language models; neuro-symbolic AI; hybrid systems; explainable artificial intelligence
Summary/Abstract: This paper explores the integration of logic programming and large language models (LLMs) in the context of the emerging field of neuro-symbolic artificial intelligence (AI) to deliver explainable and verifiable reasoning. LLMs have powerful language understanding capabilities, but they often lack formal consistency and explainability. Logic-based systems, on the other hand, offer formal reasoning, but face challenges with scalability and the processing of unstructured input data. The integration of these approaches enables the development of hybrid AI systems that combine adaptability with formal verification and explainability. This study examines the main integration strategies, reviews key systems and analyzes their potential benefits in areas requiring trust and regulatory alignment. The main technical and conceptual challenges are discussed. The findings support neuro-symbolic integration as a promising direction for developing AI systems with higher levels of interpretability and reliability, applicable to business and decision support systems.

• Details
• Contents