Knowledge Bases and User Profiling in Travel and Hospitality Recommender Systems.

Recommender systems for the travel and hospitality industries attempt to emulate offline travel agents by providing users with knowledgeable travel suggestions. The ultimate goal is to help the user in the travel planning phase trough offering a comfortable understanding of the options and also giving a select set of alternatives. This paper presents a novel approach for constructing such systems: a) creating a domain specific dialog model, b) semi-automatically building a knowledge base of ratings for the items of interest (i.e. destinations, airfare, hotel, vacation packages), and c) generating personalized recommendations ordered by relevancy. Items of interest are selected to best fit the needs of travelers, based on their individuality, interests and preferences. Explicit and tacit user feedback, as well as the extrapolation of individual user interests through attribute-based collaborative filtering, allows the system to learn rich profiles and refine its knowledge base, generating ever-improving recommendations. Empirical results confirm the hypothesis that recommender systems tend to accelerate the decision-making process by showing an improvement in look-to-buy ratios of up to 4.95 times, when compared to normal purchases on a ski travel e-commerce site. [ABSTRACT FROM AUTHOR]