On the Generation of Diverse Solution Sets in a Tourist Trip Design Problem - An Experimental Study

Route personalization is essential for designing tourist trips, but incorporating all influencing factors into mathematical models is challenging. An alternative is to obtain a diverse set of high-quality routes in advance and rank them a posteriori based on user profiles, simplifying the models and enabling straightforward solutions. This paper examines the combined use of the Modeling to Generate Alternatives (MGA) technique and route similarity measures in a Tourist Trip Design Problem (TTDP) for the generation of diverse routes. Two MGA approaches were compared to a basic evolutionary algorithm (EA) that does not prioritize diversity. Results show that while the basic EA yields solutions with higher average tourist interest, the MGA methods yield more diverse solutions with acceptable tourist interest levels. Statistically significant differences in solution diversity were found between the MGA variants, but differences were not significant in terms of the tourist interest or when different route similarity measures were considered.