Name in original language
연구지원서비스
Initiative overview
This initiative was launched to address structural limitations in research caused by vast volumes of policy research materials and a fragmented data environment, which required significant time and manpower. Previously, researchers individually collected and analysed data, resulting in similar or duplicated studies and constraints in responding to time-sensitive policy issues. Meanwhile, advances in hyper-scale AI technologies created an opportunity to redesign the entire research lifecycle and build a data-driven research ecosystem.
Accordingly, the service aims to establish an integrated research support platform that assists the full research process, from topic identification and planning to implementation, utilisation, and dissemination of results. By combining multiple large language models, a research data lake, and a retrieval-augmented generation (RAG) architecture, it provides responses grounded in reliable data sources such as policy research reports, legal information, institutional regulations, and news articles. Through this approach, it seeks to enhance research productivity, prevent duplicate studies in advance, and strengthen analytical depth and policy applicability.
Since launching a pilot service under the 2024 Hyper-scale AI-Based Public Service Development Support Programme, the foundational environment has been progressively enhanced through data standardisation, cloud infrastructure development, and the establishment of security and hallucination prevention mechanisms. In addition, operational stability in data sharing and service management has been ensured through inter-institutional coordination bodies and monitoring systems.
Results, outcomes and impacts
Currently in the pilot stage, usage remains limited in scale; however, initial user testing has confirmed the practical applicability of functions such as research topic exploration, document summarisation, and draft writing in real research workflows. In particular, the reliability of responses has been validated through a retrieval-augmented generation (RAG) structure that presents supporting source documents. The service is being continuously improved by incorporating feedback from research institution users. Through full-scale operation and expanded data integration in the future, improvements in research efficiency and reductions in similar or duplicate studies are expected.
