Scholarlink

The rapid advancement of Artificial Intelligence (AI) has prompted the vision of AI-Augmented Business Process Management Systems (ABPMSs), which aim to enhance process adaptability, explainability, and user interaction. This research proposes a hybrid framework integrating Large Language Models (LLMs) with Symbolic AI reasoners to support ABPMSs framed autonomy in a conversationally actionable way. Drawing an analogy to Kahneman’s theory of cognition, LLMs (System 1) enable intuitive, natural language interactions, while reasoners (System 2) ensure precise, rule-based decision-making and compliance with process constraints. This framework functions as a Decision Support System, explaining decisions, enabling what-if analysis, and identifying optimal trade-offs that minimize violation costs while ensuring process progression. By formalizing the frame, the framework enables precise reasoning and autonomous reframing of processes. The research also addresses fostering trust and collaboration between the ABPMS and human users. Initial results demonstrate the potential of this approach to improve decision support and dynamically align process execution with evolving constraints.

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

Integrating LLMs and Symbolic Reasoning for Framed Autonomy in AI-Augmented Business Process Management

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Angelo Casciani

摘要

The rapid advancement of Artificial Intelligence (AI) has prompted the vision of AI-Augmented Business Process Management Systems (ABPMSs), which aim to enhance process adaptability, explainability, and user interaction. This research proposes a hybrid framework integrating Large Language Models (LLMs) with Symbolic AI reasoners to support ABPMSs framed autonomy in a conversationally actionable way. Drawing an analogy to Kahneman’s theory of cognition, LLMs (System 1) enable intuitive, natural language interactions, while reasoners (System 2) ensure precise, rule-based decision-making and compliance with process constraints. This framework functions as a Decision Support System, explaining decisions, enabling what-if analysis, and identifying optimal trade-offs that minimize violation costs while ensuring process progression. By formalizing the frame, the framework enables precise reasoning and autonomous reframing of processes. The research also addresses fostering trust and collaboration between the ABPMS and human users. Initial results demonstrate the potential of this approach to improve decision support and dynamically align process execution with evolving constraints.