THOR: A Hybrid Recommender System for the Personalized Travel Experience

One of the travelers’ main challenges is that they have to spend a great effort find and choose most desired travel offer(s) among vast list non-categorized non-personalized items. Recommendation systems provide an effective way solve problem information overload. In this work, we design implement “The Hybrid Offer Ranker” (THOR), hybrid, personalized recommender system for transportation domain. THOR assigns every traveler unique contextual preference model built using solely their personal data, which makes sensitive user’s choices. This used rank offers presented each user according preferences. We reduce recommendation one binary classification predicts probability with will buy available offer. Travel are ranked computed probabilities, hence model. Moreover, tackle cold start new users, apply clustering algorithms identify groups travelers similar profiles build group. To test system’s performance, generate dataset some carefully designed rules. The results experiments show tool capable learning preferences ranks starting from those higher being selected.