Latent Dirichlet Allocation to Study Territorial Identity via In-Depth Interviews

Territories shifting from industrial to tourism-based development face complex processes of identity redefinition. Although in-depth interviews are commonly used to investigate these dynamics, their analysis often remains largely interpretative, limiting reproducibility and the systematic detection of latent themes. To fill this gap, we combine qualitative interviewing with the latent Dirichlet allocation (LDA) probabilistic topic model. By modeling each interview as a mixture of latent topics, the approach enables a data-driven extraction of recurrent themes while preserving the richness of narrative material. The results identify three main thematic dimensions: (i) cultural identity and everyday territorial belonging, (ii) tourism promotion and strategic repositioning, and (iii) development challenges and governance concerns. These topics reveal both continuity with the industrial past and tensions associated with emerging tourism-oriented futures.