본문 바로가기
  • Home

Design of a RAG-based Q&A System for University Academic Administration Assistants

  • Journal of The Korea Society of Computer and Information
  • Abbr : JKSCI
  • 2025, 30(7), pp.79~86
  • Publisher : The Korean Society Of Computer And Information
  • Research Area : Engineering > Computer Science
  • Received : June 23, 2025
  • Accepted : July 21, 2025
  • Published : July 31, 2025

Jae-Kwang Oh 1 Soo Kyun Kim 1

1제주대학교

Accredited

ABSTRACT

This paper aims to implement a Retrieval-Augmented Generation based question-answering system to efficiently process and utilize various unstructured academic administrative documents, such as university regulations, course catalogs, and admissions guidelines. To achieve this, a document parsing and preprocessing pipeline was designed to accurately interpret the structure of PDF and HWP formats, enabling generative AI models to comprehend context and generate natural responses. To overcome the limitations of large language models (LLMs), such as token constraints and vector-based similarity search inefficiencies, optimal chunking and embedding strategies were developed. Furthermore, a hybrid retrieval method was integrated using the LangChain framework to improve response accuracy and reliability. The proposed system provides university members with fast and precise access to institutional information, enhancing administrative efficiency and offering a practical reference for future AI-driven knowledge retrieval systems in higher education and public institutions.

Citation status

* References for papers published after 2024 are currently being built.