A comprehensive deep learning approach for topic discovering and sentiment analysis of textual information in tourism

Abstract

Automatic discovery of underlying themes in a document collection is a valuable task across many disciplines. Advanced techniques can be challenging for non-experts in data science to understand. To address these challenges, this work proposes a comprehensive deep-learning-based method for gathering, preprocessing, analyzing, and classifying text data to discover topics in extensive collections of documents. This method produces results understandable to humans, which is especially valuable in fields outside of data science. We tested the proposed method on a corpus of all news articles (in English) from the USA and Canada about Cancun, a popular tourist destination in Mexico, published between July 2021 and July 2022. Despite negative media coverage, we discovered a positive attitude toward Cancun’s amenities. This information can help destination management organizations monitor the destination’s digital reputation and design effective communication campaigns for potential visitors who consult these sources of information.

Keywords

• Destination image
• Online news articles
• Deep learning
• Topic modeling
• Mexico