Integrating Large Language Models into Decision Support Systems for Objective Student Competition Participant Selection

Eka V. Dangkua (Universitas Negeri Gorontalo)
Fazrul Anugrah Sahi (Universitas Negeri Gorontalo)
Nikmasari Pakaya (Universitas Negeri Gorontalo)
Indhitya R. Padiku (Universitas Negeri Gorontalo)
Muchlis Polin (Universitas Negeri Gorontalo)
Rahmat Taufik R. L Bau (Universitas Negeri Gorontalo)

Journal of Renewable Engineering
Vol. 3 No. 2 (2026): JORE - April

The selection process for competition participants in the Informatics Engineering Department at Gorontalo State University is still carried out manually, potentially leading to subjectivity and suboptimal identification of deserving students. This study developed an artificial intelligence-based selection system using a Large Language Model (LLM) integrated with the Retrieval-Augmented Generation (RAG) approach. The developed system, named Scout, recommends competitions that match student profiles based on academic data, interests, experience, and achievements. The system evaluation used the Precision@K and Hit@K metrics to measure recommendation accuracy, and RAGAS to assess the quality of retrieval and chatbot responses. Test results showed that the Scout system obtained a Precision@3 score of 0.87 and a Faithfulness score of 0.91, indicating high recommendation relevance and factual consistency. Thus, the implementation of LLM and RAG has proven effective in increasing the objectivity and efficiency of the selection process and has the potential to become the basis for the development of an AI-based academic decision support system in higher education