Smart Search and Chatbot for Real Estate and Construction Data

Overview

To resolve the inefficiency of processing 10,000+ annual building-related complaints, the Korea Real Estate Board developed a SearchGPT/SmartChat-based pilot service to reduce staff workload.

Project description

For the service pilot, approximately 900 complaint cases, 100 pages of the Building Act, and 400 pages of daily life regulations were used as training data.

For model training, the Korea Real Estate Board preprocessed its complaint and legal data and reviewed the query generation results from the SmartChat system. Subsequently, the preprocessed data was trained and indexed in SearchGPT/SmartChat, ultimately leading to the development of a testable demo page.

Key services include a SmartChat-based chatbot service for complaints and construction-related information, and a SearchGPT-based information retrieval function for various laws and regulations or complaint responses.

Key technological features

• LLM Training Dataset: Preprocessing of Civil Complaints and Legal Statute Data Held by the Korea Real Estate Board
• SearchGPT/SmartChat-Based Service: Training and Indexing Preprocessed Data on SearchGPT, and Generating Queries Based on SmartChat

Expected outcomes

The service reduces the burden of civil petitions by providing easy access to historical data through integrated search and chatbot functions. Following the successful PoC, we plan to implement a full-scale service featuring building regulation updates and an advanced similar-petition search function for the upcoming year.

Launch year: 2023

Attribution

Case study courtesy of the National Information Society Agency, Republic of Korea, Apolitical's Content Partner. Sourced from Use Cases of Public AI Service, Vol.2, https://eng.nia.or.kr/site/nia_eng/ex/bbs/View.do?cbIdx=31975&bcIdx=29182&parentSeq=29189