基于大语言模型的企业异构数据融合查询  

Enterprise Heterogeneous Data Fusion Query Based on Large Language Model

作　　者：吴春龙 汪敏 陈智超 WU Chunlong;WANG Min;CHEN Zhichao

机构地区：[1]商飞智能技术有限公司,上海201210 [2]上海飞机制造有限公司,上海201323

出　　处：《科技创新与应用》2025年第10期1-5,共5页Technology Innovation and Application

摘　　要：大语言模型在智能问答、文本生成、语言翻译、辅助编程等创造性的场景应用十分广泛,但是在需求精确性的场景下应用却受到诸多限制。该文主要研究采用大语言模型,在知识图谱和向量知识库的加持下,结合Prompt提示工程、微调、LangChain等技术,融合结构化数据和非结构化数据,实现在限定知识范围内的精确查询,探索大语言模型应用的新方式。Large Language Models(LLMs),also known as big models,have extensive applications in creative scenarios such as intelligent question answering,text generation,language translation,and programming assistance.However,their application in precision-demanding contexts is often subject to various limitations.This paper primarily investigates the utilization of large language models,coupled with the support of Knowledge Graphs and Vector Knowledge Bases.By incorporating techniques such as Prompt Engineering,fine-tuning,and LangChain,we aim to fuse structured and unstructured data and achieve precise queries within a defined knowledge scope.This research explores new approaches for the application of large language models.

关 键 词：大语言模型 异构数据 知识图谱 向量知识库 融合查询 

分 类 号：TP312[自动化与计算机技术—计算机软件与理论]