摘要
本文通过文献调研方法梳理了人工智能(AI)的发展历史和AI大模型诞生的技术背景,归纳了AI大模型训练的技术方法;通过调研当前国内外主要的AI大模型基本情况,参考AI大模型的主流观点,提出AI大模型发展的基本态势。针对当前AI大模型的能力上限和边界、存在的问题和面临的挑战,借鉴AI大模型行业落地途径及部分应用案例,提出了AI赋能不动产登记的基本思路,即在自然资源行业大模型中外挂不动产向量知识库和不动产知识图谱,通过行业大模型、向量库和知识图谱的协同与优势互补构建不动产登记AI能力底座,再通过AI助手将AI能力嵌入不动产登记系统,实现AI赋能不动产登记全流程、各环节。
This paper article reviews the development history of artificial intelligence(AI)and the technical background of the emergence of AI large models through literature research methods,and summarizes the technical methods to train AI large models.By investigating the basic situation of the main AI large models currently available both domestically and internationally and referring to the mainstream view of AI foundation models,the basic development situation of AI large models is proposed.In response to the capacity limitations and boundaries,existing problems and challenges of the current AI large models,and by referring to the implementation methods and some application cases of AI large models in the industry,the basic idea of AI empowering real estate registration is proposed in this paper.The basic idea is to attach real estate vector knowledge base and real estate knowledge graph to the foundation models of the natural resources industry,construct the AI capacity base of real estate registration through the collaboration and complementary advantages of the industry foundation models,vector libraries and knowledge graphs,and then embed the AI capacity into the real estate registration system by AI assistants so as to realize the full process and various links of AI empowering real estate registration.
作者
刘聚海
胡玥
姜喆
贾文珏
程立海
刘天羽
孙华清
李彦
张菲菲
LIU Juhai;HU Yue;JIANG Zhe;JIA Wenjue;CHENG Lihai;LIU Tianyu;SUN Huaqing;LI Yan;ZHANG Feifei(Information Center of Ministry of Natural Resources,Beijing 100812,China)
出处
《自然资源信息化》
2024年第4期1-18,共18页
Natural Resources Informatization
关键词
人工智能
AI大模型
机器学习
深度学习
自然资源
不动产登记
综述
artificial intelligence
AI large models
machine learning
deep learning
natural resources
real estate registration
overview