AI Bridges Natural Language and Answer Set Programming with Hybrid LLM and AMR Approach

A team of researchers has introduced a new method that bridges the gap between natural language and Answer Set Programming (ASP) using a hybrid approach involving Large Language Models (LLMs) and Abstract Meaning Representation (AMR) parsing. This innovation seeks to make ASP, a declarative programming paradigm based on logic programming and non-monotonic reasoning, more accessible to individuals without extensive programming knowledge. The proposed system leverages LLMs to simplify natural language sentences, identify keywords, and generate simple facts. Subsequently, AMR graphs are parsed from the simplified language to systematically generate ASP constraints. By minimizing the role of LLMs and focusing on AMR parsing, the system aims to create a more explainable and lightweight solution for converting natural language into code that can solve complex logic problems. The system successfully generates entire ASP programs capable of solving combinatorial logic puzzles. This hybrid approach represents a significant stride towards developing tools that are easier to use, more robust, and more flexible for users who are not experts in the field of Knowledge Representation and Reasoning (KRR). Future work will focus on expanding the capabilities of the system to handle more complex logic constructs and improve overall consistency.