Optimizing the Regulatory Information Search System

Overview

As the business environment becomes more sophisticated and government regulations grow increasingly complex and diverse, there is a rising demand for a systematic search infrastructure that allows corporations and the public to access regulatory information with ease. This project was planned to enhance the information accessibility and utility of the existing Regulatory Reform Portal (www.better.go.kr) and the National Law Information System.

Project description

The GBC export support platform 'gbcprime' was implemented to provide export consultation chatbots, overseas market information, customized buyer recommendations, best practice guidance, and information on related GBC programs and support initiatives. Considering its Proof of Concept (PoC) status, a representative test utilizing the GBC Prime LA Office database (2022–2024) demonstrated suitability with a 0% error rate across 183 cases involving 50 concurrent users.

Key technological features

• Optimization and Reinforcement Learning based on User Feedback
• Korean LLM-based Fine-tuning
• Utilization of X2BEE Base Models: AI Chatbot, AI Search, AI Customer Service Analytics, AI Marketing Copywriter, etc.

Expected outcomes

Through Agile-based LLM development and fine-tuning, prediction uncertainty was reduced by 21.9%, and the user follow-up question rate decreased by 24.6%. By enabling corporations and the public to access regulatory information quickly and accurately, this project has improved regulatory compliance rates and contributed significantly to enhancing administrative efficiency.

Launch year: 2024

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=29209