Discovering tourist typologies through photo mining: A hybrid approach

In this thesis, 13,982 digital photographs collected from destinations in the Mediterranean regions were analysed in order to understand tourist behaviour in the digital age from the perspective of the "e-mediated tourist gaze" and to identify different tourist typologies. The photographs were first labelled using the Blip-2 visual language model. Subsequently, five distinct and recognisable tourist segments were identified using NMF topic modelling based on label sizes. To typologically define these segments, sample photographs representing each cluster with the highest scores were presented to 17 experts for review and interpreted using Barthes' semiotic theory. Similarly, using an AI tool with visual analysis capabilities (genAI), the best representative photos were thematised at the typological level by creating prompts using a semiotic approach. The resulting five original tourist typologies are, in order, "Cultural Explorers (tourists focused on historical and cultural heritage, seeking meaning)", "Nature and Peace Seekers (those who prioritise untouched natural beauty and tranquillity, escaping crowds)", "Connection Travellers (those who place the travel process and means of transportation at the centre of their experience)", "Social Interactors (those who aim to establish authentic social connections with the local population and other people)," and "Macro Observers (those who observe the places they visit from a panoramic and holistic perspective)." Statistical analyses (Chi-Square and Goodness-of-Fit Analysis) have revealed a meaningful relationship between these tourist typologies and destinations. For example, Cultural Explorers are more commonly found in historical regions such as Crete and Rhodes, while Nature and Peace Seekers are strongly associated with natural areas like Fethiye and Kaş. Thus, this thesis study has provided insights into the different motivations of postmodern tourists with the help of extensive visual data analysis. It has also shown that these typologies are not rigid patterns and that destinations in similar regions may have their own unique tourist profiles. It is believed that the typologies that have emerged will help to understand classic tourist typologies in today's conditions, beyond mere behavioural segmentation.