Chatbot Interaktif Berbasis Transformer untuk Penyediaan Informasi Dinamis di Industri Esports

Calvin, Calvin
Mawardi, Viny Christanti
Lewenusa, Irvan

Communication and Information Journal
Vol. 3 No. 2 (2025): Comit: Communication and Information Journal

The esports industry is experiencing rapid growth and has given rise to the need for an information system that is able to present news quickly and relevantly. This study develops an interactive chatbot based on the BART (Bidirectional and Auto-Regressive Transformers) model to answer user questions directly based on information collected through web scraping. The system is designed using Python and Flask, and uses MongoDB as data storage. Evaluation is carried out through F1-Score and Conversation-Based Evaluation to assess system performance in terms of relevance, coherence, and fluency of answers. The results show that the chatbot is able to provide accurate and coherent answers with fast response times, although there are still challenges in terms of recall and understanding ambiguous contexts.