Development of Trade and Investment GPT Service Based on Overseas Country and Certification Information

Overview

Existing trade and investment platforms operate independently, limiting data linkage and analysis capabilities. Accordingly, this project proceeded with the goal of establishing a Trade and Investment GPT to integrate country-specific trade and investment information and create a foundation that anyone can easily utilize.

Project description

KOTRA collected and integrated its country-specific trade information, investment information, and overseas certification data. Based on this, it built a QA dataset and RAG model for AI training. The collected trade information includes trade agreements, promising export items, and logistics systems for 96 countries, while the investment information reflects each country's investment trends and important considerations. Additionally, overseas certification information was refined and processed, encompassing data on over 1,180 certification procedures, agencies, costs, and required durations.

Based on this data, this project developed an AI-based trade and investment GPT model and implemented it to provide accurate information in real time through a chat service utilizing RAG technology. The developed service was designed to be linked with KOTRA's TriBIG (Trade and Investment Big Data Platform) for public access, enabling businesses and citizens to more easily utilize overseas market information.

Key technological features

• Building Datasets by Refining and Processing Source Data in Various Forms (PDF/Text/Tables, etc.) on Webpages Using Topic-Based Chunking Strategies
• Building RAG (Retrieval-Augmented Generation)-based models using on-premises proprietary LLM and external API integration
• Performing model-specific response comparisons using APIs such as GPT-4o, HCX-003, and SoLAR-MINI

Expected outcomes

The establishment of the Trade and Investment GPT will enable users to view previously scattered overseas market information at a glance, significantly improving access to such information for businesses and citizens. In particular, companies preparing for exports will reduce the burden of individually searching for information and can establish more effective market entry strategies by utilizing AI-based customized information. Furthermore, it is expected to provide practical assistance by allowing companies preparing for overseas expansion to easily grasp investment information.

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.1,  https://eng.nia.or.kr/site/nia_eng/ex/bbs/View.do?cbIdx=31975&bcIdx=28997&parentSeq=29018