Enhancing the Accuracy of Question-Answering Systems Using Machine Learning: A Case Study in the Tourism Domain

With the rapid development of smart tourism, the accuracy of question-answering systems (QAS) has become crucial in enhancing user experience, especially in contexts where human-computer interaction is central. In tourism, QAS typically handle queries related to locations, attractions, and transportation, where the information varies widely and the phrasing can change frequently. This makes the semantic understanding and response generation a challenging task for QAS. This paper explores how machine learning (ML) has been employed to improve QAS accuracy, focusing on intent recognition, semantic matching, context modeling, and the use of pre-trained language models (PLMs). By analyzing the various techniques and methods used in tourism, this study provides a comparative overview of their effectiveness in addressing domain-specific needs. The findings suggest that incorporating domain knowledge, fine-tuning models, and leveraging multimodal semantics and multilingual capabilities are key strategies for improving the accuracy of tourism-specific QAS.