Big data analytics of online news to explore destination image using a comprehensive deep-learning approach: a case from Mexico

Destination image has been a subject of great interest to tourism scholars for several decades. Since the nature of this social construct is highly dynamic, its study poses new challenges under the current conditions of contemporary tourism practices. Considering that the image formation process can be influenced positively or negatively by multiple sources of information available to individuals, it is surprising that analyses of autonomous formation agents, such as online news, have received limited attention in related literature. Although existing studies have explored the influence of this information on image formation, intention to visit, and actual behavior, these normally adopt traditional methodologies to collect information, circumscribing the analysis to limited samples. The main objective of this work is to propose an innovative automated approach based on deep learning aimed at collecting and analyzing available textual data on the internet, such as online news, to produce a more comprehensive picture of the destination image in these sources of information. In order to test this approach, a destination from the country of Mexico was selected as a case study: Cancun. Given that the USA and Canada represent almost 60 percent of all international visitors to Mexico, the information search focused on this geographical context. A total of 3845 online news making reference to Cancun were retrieved during an entire year (July 2021-2022). The analysis of this information allowed the identification of recurrent topics covered by the media in both countries regarding destination safety issues, criminal activities, and the evolution of travel restrictions due to the COVID-19 pandemic. In addition to these topics, favorable coverage could also be detected including topics such as existing amenities in all-inclusive resorts as well as the recognition of Cancun as an ideal tourist destination for the international traveler. In practical terms, we believe this information can be useful for local government and DMOs to explore the evolution of the destination's image as well as to identify sensitive issues covered in the media that require the implementation of communication strategies to counteract any potential negative effect. Finally, the proposed approach effectively contributes to making the tasks of destination image evaluation easier and faster than traditional research strategies.