Developing Tourism Users’ Profiles with Data-Driven Explicit Information

In recommender systems (RSs), explicit information is often preferred over implicit because it much more accurate than or predicted information; for example, the user can enter about his interests directly into system, and system will generate recommendations him. Receiving information, however, may be difficult a system. Explicit demographic might uncomfortable some users, extremely common questions, such as race, gender, income, age, lead to bias unfair recommendations. As result, in this study, we propose method, which collected from new does not contain enquired data driven. Users’ interest tourism activities used identify seven categories of tourism. The mapping between extracted established with multilabel classification (MLC) algorithm. user’s 18 by rating only categories. Common MLC algorithms different classifiers were evaluate proposed method. best result relates binary relevance Naïve Bayes classifier, also outperforms entitled collaborative filtering (CF) baseline models. method capture users’ develop their profiles without receiving information. Also, compared CF, addition slight advantage metrics, requires ratings predict activities. contrast, CF require at least 15 ng records unknown (3-4 activities) achieve performance close