Scholarlink

<p>In large-scale digital service networks, interactions and coordination among diverse stakeholders are highly complex. First, we propose the TRGPS (Topological Role-Goal-Process-Service) meta-model to represent the requirement model of digital service networks. The rapid evolution of digital service networks often leads to inconsistencies between existing services and newly integrated ones. Existing goal-based modeling approaches often struggle to identify conflicts among stakeholders’ objectives, making effective conflict resolution difficult. To address this issue, we propose an automatic conflict detection and resolution approach for digital service network requirements modeling, leveraging Large Language Models (LLMs) and the TRGPS meta-model. We define a set of rules for detecting goal conflicts and use Chain of Thought (CoT) prompting, a state-of-the-art (SOTA) prompting technique, as input to LLMs. Our approach enables LLMs to reason about potential conflicts between goal elements and relationships in digital service networks and generate recommendations for resolution. We also develop a tool that supports visual identification and resolution of goal conflicts, assisting requirements analysts in their modeling tasks. Qualitative experiments and case studies confirm the effectiveness of the proposed approach.</p>

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

Automatic conflict detection and resolution in digital service network requirement models using large language models

  • 期刊论文

  • 发表日期：

    2025年12月5日

查看全文
  • Siyu Nan,
  • Yu Qiao,
  • Yifan Xie,
  • Zilong Zhang,
  • Yaling Luo,
  • Bing Li,
  • Jian Wang

摘要

In large-scale digital service networks, interactions and coordination among diverse stakeholders are highly complex. First, we propose the TRGPS (Topological Role-Goal-Process-Service) meta-model to represent the requirement model of digital service networks. The rapid evolution of digital service networks often leads to inconsistencies between existing services and newly integrated ones. Existing goal-based modeling approaches often struggle to identify conflicts among stakeholders’ objectives, making effective conflict resolution difficult. To address this issue, we propose an automatic conflict detection and resolution approach for digital service network requirements modeling, leveraging Large Language Models (LLMs) and the TRGPS meta-model. We define a set of rules for detecting goal conflicts and use Chain of Thought (CoT) prompting, a state-of-the-art (SOTA) prompting technique, as input to LLMs. Our approach enables LLMs to reason about potential conflicts between goal elements and relationships in digital service networks and generate recommendations for resolution. We also develop a tool that supports visual identification and resolution of goal conflicts, assisting requirements analysts in their modeling tasks. Qualitative experiments and case studies confirm the effectiveness of the proposed approach.