Scholarlink

This study investigates the automation of decision logic representation through AI-based methods, specifically focusing on the generation of Decision Model and Notation (DMN) tables using Large Language Models (LLMs) enhanced with Retrieval-Augmented Generation (RAG). The research evaluates the feasibility of replacing traditional DMN generation methods with LLM-generated tables, ensuring correctness in structured expressions and handling complex logic. The experiment involved prompting Anthropic AI Claude Sonnet 3.5 with structured business rules and validating its output. Results indicate that while LLMs can generate structured DMN tables with RAG support, improvements in prompt engineering and dataset precision are necessary to mitigate hallucinations.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

AI-Based Efficient Automation of Decision Logic Representation

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Olga Cherednichenko,
  • Vladyslav Maliarenko

摘要

This study investigates the automation of decision logic representation through AI-based methods, specifically focusing on the generation of Decision Model and Notation (DMN) tables using Large Language Models (LLMs) enhanced with Retrieval-Augmented Generation (RAG). The research evaluates the feasibility of replacing traditional DMN generation methods with LLM-generated tables, ensuring correctness in structured expressions and handling complex logic. The experiment involved prompting Anthropic AI Claude Sonnet 3.5 with structured business rules and validating its output. Results indicate that while LLMs can generate structured DMN tables with RAG support, improvements in prompt engineering and dataset precision are necessary to mitigate hallucinations.