<![CDATA[Despite significant advancements in deep learning models for chatbots, comprehensive analyses tailored to the tourism sector remain limited. This study addresses the gap by comparing the performance of six prominent models—MLP, RNN, GRU, LSTM, BiLSTM, and CNN—in creating chatbots designed to address traveler needs such as information about facilities, ticket prices, activity suggestions, and operational details. The methodology includes key stages such as data collection, preparation, model training, and evaluation using accuracy, precision, recall, F1-score, and qualitative assessments. The dataset, derived from interviews with managers of 11 tourism destinations, captures critical details to replicate real-world user interactions. The results indicate that the CNN model performed the best, achieving the highest accuracy (0.98), precision (0.99), recall (0.98), and F1-score (0.98), showcasing its ability to effectively handle user queries by identifying relevant patterns in data. While MLP achieved strong accuracy (0.94), its simpler design limited its capacity to manage complex questions. The RNN model had the lowest accuracy (0.82), highlighting its challenges in understanding structured information. These findings confirm CNN as the most effective model for retrieval-based chatbots in tourism, balancing accuracy and practicality. This research offers valuable insights for improving AI-driven tourism tools, providing guidelines for selecting optimal models and enhancing chatbot performance to enrich the traveler experience.]]>
Copyrights © 2025
