<![CDATA[Information technology plays a crucial role in enhancing knowledge and awareness, particularly in educational institutions with intense competition among universities. Students and the general public still face difficulties in obtaining the necessary information due to the numerous available platforms and the lack of effective search features. This research aims to develop an AI-based chatbot, specifically using Large Language Models (LLM), that can offer efficient and responsive solutions for information services in educational institutions. By integrating the chatbot into the Telegram application, it is expected to facilitate users in quickly and accurately obtaining information. This chatbot is built with a Retrieval Augmented Generation (RAG) approach that enhances LLM capabilities and helps in obtaining fast and accurate information. The chatbot is developed using software such as Langchain to manage the RAG process, Python as the programming language, ChromaDB as the vector database, and Gemini AI to support the LLM model. Performance evaluation is conducted using RAGAS to ensure the quality and accuracy of responses. Testing with 10 sample questions showed positive results, a context precision score of 93%, faithfulness 100%, answer relevancy 96%, context recall 100%, answer correctness 71%, and answer similarity 93%. These results indicate that an RAG-based chatbot can be an effective tool to improve information accessibility in educational institutions.]]>
Copyrights © 2024
