What is of interest for tourists in an alpine destination: personalized recommendations for daily activities based on view data

• Published in: Journal of ambient intelligence and humanized computing 2020-11, Vol.11 (11), p.4545-4556
• Main Authors: Majeed, Tahir, Stämpfli, Aline, Liebrich, Andreas, Meier, René
• Format: Article
• Language: English
• Subjects: Applications programs; Artificial Intelligence; Collaboration; Computational Intelligence; Customization; Engineering; Holidays & special occasions; Hybrid systems; Information overload; Internet; Mobile computing; Original Research; Recommender systems; Robotics and Automation; Smartphones; Tourism; Tourist attractions; Travel; User Interfaces and Human Computer Interaction
• ISSN: 1868-5137; 1868-5145
• DOI: 10.1007/s12652-019-01619-1

Smartphones are nowadays important tools for tourists. For instance, while on the go in a destination, tourists can use smartphones to find places of interest and to identify activities that might be of interest, as well as a wealth of related information and even special offers in real-time. However, it is time consuming and not easy for tourists in an unknown destination to choose among the numerous options available. Recommendations for instance from other tourists with similar interests would help immensely. Indeed, places and activities are often reviewed and rated by other tourists, however, this information is typically not personalized. This article proposes a recommender system as part of an evolving mobile destination app. Our recommender app is capable of providing personalized recommendations to tourists thereby facilitating and enriching tourists’ experience and stay. This work is based on two qualitative studies towards exploring the information needs of tourists in an alpine destination. These studies were conducted using the mobile ethnography approach and semi-structured interviews. A hybrid recommender system is proposed that uses implicit user feedback in the form of view duration. The proposed system was tested using real data derived from tourists using the mobile app in a Swiss alpine destination. The results of these experiments demonstrate that the system is capable of providing high-quality and diverse recommendations. The core contribution of this work lies in the transformation of the viewing durations to a set of preference values and in learning the optimized weights of the parameters of a hybrid system utilizing an energy minimization framework.