PANAS-TDL: A Psychrometric Deep Learning Model for Characterizing Sentiments of Tourists Against the COVID-19 Pandemic on Twitter

Peña, Alejandro; Mesias, Jorge; Patiño, Alejandro; Carvalho, Joao Vidal; Gomez, Gregorio; Ibarra, Kevin; Bedoya, Santiago

One of the main sector that moves the economy of the countries worldwide is tourism and its associated services. Dynamics like globalization has led these countries to create tourism services with global standards, however, in the context of COVID-19 pandemic, these services have been affected as shown the social networks. This fact led to a change in the perception of tourists against a destination. In order to unify this change in an objective manner, we propose a Deep Learning model that integrates a PANAS scale (Positive and Negative Affect Scale) (PANAS-tDL), to characterize a tourist destination based on a series of potential factors (weather conditions, healthy, holidays, seasonality and economic factors) identified in comments obtained from a social network like Twitter. The results obtained by the PANAS-tDL model show its good performance evaluating the change of perception of tourists against four destinations affected by COVID-19, taking as reference the 11-sentiment scale defined by PANAS-t scale. Thanks to adaptation capacity, the model can be extended to evaluate the change in perception of tourists using different social networks and to evaluate different marketing strategies to promote a destination.