<![CDATA[This study aims to develop a Retrieval-Augmented Generation (RAG)-based academic chatbot to provide accurate, relevant, and official document-based academic information for students of the Informatics Engineering Study Program, FTI UNISSULA. The methods used include collecting and pre-processing academic documents, chunking processes, forming vector representations using the Sentence-BERT model, and storing them in the FAISS database to support semantic search. The RAG system integrates document retrieval results with the capabilities of the Large Language Model (LLM) in generating contextual responses. System evaluation was carried out using the ROUGE-1 and BLEU-4 metrics on 50 questions consisting of FAQ and non-FAQ categories. The test results showed that the system was able to respond to all questions given, with high performance in the FAQ category (ROUGE-1 of 0.957 and BLEU-4 of 0.877), and lower performance in the non-FAQ category due to paraphrasing variations in academic documents. These results indicate that the RAG approach is effective in improving the accuracy and relevance of academic chatbot answers, and is able to reduce the risk of misinformation compared to a purely generative approach.]]>
Copyrights © 2026
