IMPLEMENTATION OF AN NLP-BASED CHATBOT FOR CAVE DESCRIPTIONS IN MAROS REGENCY

Abstract

Maros Regency is known for the Maros–Pangkep Karst area, which contains caves with high historical, cultural, and geological value. However, information about these caves is still scattered and not yet integrated into a single platform that is easily accessible. This study designs and implements a chatbot based on Natural Language Processing (NLP) to provide automated and interactive descriptions of the caves.The system utilizes the DistilBERT model for natural language processing and Supabase as a cloud-based database. Three caves that serve as the focus of this research are Gua Kharisma, Gua Patta, and Gua Leang Sampeang Dua. The dataset consists of question–answer pairs based on five criteria: history, ornaments, size, location, and distinctive features of the caves.Black-box testing shows that the system functions according to the design, providing fast, accurate, and relevant responses.