摘要
为解决钢铁企业数智化实施过程中存在的信息透明度不足、传递不及时、数据难以分析以及分析结果难以被业务人员理解等问题,引入知识图谱和大语言模型技术,基于关系型数据库、非结构化文档等多来源数据,构建钢铁制造相关管理指标知识图谱。在此基础之上,通过检索增强生成技术,以管理指标知识图谱作为内部数据来源和推理引擎,大语言模型作为自然语言理解和生成引擎,构建业务用户友好的钢铁制造管理问答应用,为其提供及时准确的生产、管理决策参考。
We introduce knowledge graphs and large language models,to provide solutions to challenges during the implementation of digital intelligence in steel-making companies,including lack of transparency,delayed communication,and difficulty in understanding complicated analysis results for management personnel.Based on big data from multiple sources like relational databases and unstructured documents,we construct a steel manufacturing management indicator system in the form of a knowledge graph.We then utilize Retrieval Augmented Generation(RAG),to develop a user-friendly Q&A application on steel-manufacturing topics,with the aforementioned indicator system serving as a data source and reasoning engine,and the large language model serving as a natural language understanding and text-generation engine.The application provides users with timely and accurate references to their decision-makings on manufacturing managementt.
作者
李川阳
张洪
Li Chuanyang;Zhang Hong(Xinjiang Bayi Iron and Steel Co.,Ltd.,Urumqi 830063,Xinjiang;Shanghai Baosight Software Co.,Ltd.,Shanghai 201900)
出处
《武汉工程职业技术学院学报》
2024年第2期22-27,共6页
Journal of Wuhan Engineering Institute