Geo-localized, mobile applications can simplify a tourist visit, making the relevant Point of Interests more easily and promptly discernible to users. At the same time, such solutions must avoid creating unfitting or rigid user profiles that impoverish the users' options instead of refining them. Currently, user profiles in recommender systems rely on dimensions whose relevance to the user is more often presumed than empirically defined. To avoid this drawback, we build our recommendation system in a two-step process, where profile parameters are evaluated preliminarily and separately from the recommendations themselves. We describe this two-step evaluation process including an initial survey (N=206), and a subsequent controlled study (N=24). We conclude by emphasizing the benefit and generalizability of the approach.
|Titolo della pubblicazione ospite||Proceedings of the 12th Biannual Conference on Italian SIGCHI - CHItaly '17|
|Numero di pagine||5|
|Stato di pubblicazione||Published - 2017|
|Nome||ACM International Conference Proceeding Series|
- Human-Computer Interaction
- Computer Vision and Pattern Recognition
- Computer Networks and Communications