Application of Large Language Model for New Student Admission Chatbot

Anwar, Rafidan
Pratiwi, Heny
Wahyuni, W

IJISTECH
Vol 8, No 6 (2025): The April edition

This study aims to develop a chatbot system based on a Large Language Model (LLM) that provides information related to new student admission in higher education. The system utilizes the SentenceTransformer model to generate embeddings of question and answer texts, as well as FAISS for vector-based search. Additionally, LLAMA is used to generate context-based answers, allowing the chatbot to provide more dynamic and relevant responses. System evaluation is conducted using ROUGE-1, ROUGE-2, and ROUGE-L metrics. The evaluation results show an average ROUGE-1 Precision of 54.89%, ROUGE-2 Precision of 47.37%, and ROUGE-L Precision of 52.72%. The Recall scores for ROUGE-1, ROUGE-2, and ROUGE-L are 89.43%, 74.08%, and 82.91%, respectively