| 논문명 |
대규모 언어모델(LLM)을 활용한 건축민원 대응 효율화 방안 연구 / A Study on the Efficient Response to Architectural Civil Complaints Using Large Language Models(LLM) |
| 저자명 |
조상규(Cho, Sang-Kyu) ; 김신성(Kim, Shin-Sung) |
| 수록사항 |
대한건축학회논문집, Vol.40 No.9 (2024-09) |
| 주제어 |
인공지능;대규모 언어모델;건축법;질의응답;법령해석제도;벡터데이터베이스 ; Artificial Intelligence;Large Language Model;Architectural Law;Legal Interpretation System;Vector Database |
| 요약1 |
이 연구는 건축 법규의 복잡성과 행정적 부담을 줄이기 위해 대규모 언어 모델을 사용한 SPARC 모델을 개발했다. 국토교통부의 자료를 활용하여 법적 해석의 효율성을 높였으며, 시스템은 일반 문의에 대해 80% 이상, 복잡한 사례에 대해서는 70%에서 100%의 정확도를 달성했다. 이는 규제 행정에 AI를 적용한 첫 사례이다. |
| 요약2 |
This study addresses the complexity of architectural laws and regulations and their administrative burden, focusing on improving efficiency in
the interpretation and query-response processes using large-scale language models. The research centers around the development and
implementation of the SPARC (Semantic Processing for Architecture Regulation Compliance) engine, primarily utilizing data from inquiries
and complaints submitted to the Ministry of Land, Infrastructure, and Transport regarding architectural laws. This prototype system is
designed to augment reference information necessary for legal interpretation, and its effectiveness was validated through a quality assessment
of system responses to actual complaint data. The results show that the system achieved an accuracy rate of over 80% for general inquiry
complaints with clear conclusions and 70% to 100% for more complex cases requiring legal interpretation by the legislative affairs office.
This research represents the first attempt to apply AI in the field of regulatory administration, providing a critical technical and policy
foundation for the development and operation of AI-based systems for interpreting architectural regulations. |
