Stock movement prediction in a hotel with multimodality and spatio-temporal features during the Covid-19 pandemic

The COVID-19 pandemic has underscored the importance of accurate stock prediction in the tourism industry, particularly for hotels. Despite the growing interest in leveraging consumer reviews for stock performance forecasting, existing methods often need to integrate the rich, multimodal data from these reviews fully. This study addresses this gap by developing a novel deep learning model, the Multimodal Spatio-Temporal Graph Convolutional Neural Network (MSGCN), specifically designed to predict hotel stock performance. Unlike traditional models, MSGCN captures the spatial relationships between hotels using a graph convolutional network and integrates multimodal information-including text, images, and ratings from consumer reviews-into the prediction process. Our research builds on existing literature by validating the efficacy of multimodal data in improving stock prediction and introducing a spatio-temporal component that enhances prediction accuracy. Through rigorous testing on two diverse datasets, our model demonstrates superior performance compared to existing approaches, showing robustness during and after the COVID-19 pandemic. The findings provide valuable insights for hotel managers and consumers, offering a powerful tool for making informed business decisions in a rapidly evolving market.