基于大模型的不动产登记智能问答方法  

Intelligent question&answering methods for real estate registration based on large language models

作　　者：董承玮[1,2] 李云汉 邢晨 刘世凡 DONG Chengwei;LI Yunhan;XING Chen;LIU Shifan(Beijing Institute of Surveying and Mapping,Beijing 100038,China;Beijing Key Laboratory of Urban Spatial Information Engineering,Beijing 100038,China)

机构地区：[1]北京市测绘设计研究院,北京100038 [2]城市空间信息工程北京市重点实验室,北京100038

出　　处：《北京测绘》2025年第3期253-258,共6页Beijing Surveying and Mapping

基　　金：北京市自然科学基金(4214069)。

摘　　要：为探索如何在不动产登记业务中提供智能问答服务,本文提出一种基于大语言模型的不动产政策智能问答方法,并结合不动产政策知识库与信息检索提升问答准确性。该方法首先通过搜索引擎和语义检索从政策知识库得到政策文本候选集合;然后利用排序模型进行排序,得到与问题最相关的候选政策文本;最后基于问题、政策文本构建提示(Prompt)并输入大模型生成答案。相较于传统方法,该方法在确保答案专业性和准确性的同时,提供了更灵活、自然的自助问答服务,对构建不动产登记智慧问答系统具有一定的参考意义。To explore the feasibility of applying intelligent question&answering(Q&A)services in real estate registration,this paper proposed an intelligent Q&A method for real estate policy based on large language models and adopted a real estate policy knowledge base and information retrieval to enhance Q&A accuracy.This method first obtained a candidate set of policy texts from the policy knowledge base through search engines and semantic retrieval.It then used a ranking model to sort and identify the most relevant candidate policy texts to the question.Based on the question and the policy texts,a prompt was constructed and fed into the large language model to generate answers.Compared to traditional methods,this method ensured the professionalism and accuracy of the answers while providing a more flexible and natural self-service Q&A,which has significant reference value for building intelligent Q&A systems in real estate registration.

关 键 词：不动产登记 智能问答 知识库 语义检索 大模型 

分 类 号：P258[天文地球—测绘科学与技术]