On-Tour Interactive Travel Recommendations

Travelers who access the Internet through currently available web information systems often experience difficulty in selecting interesting products. This is especially true for on-the-move travelers who browse information repositories using mobile devices with poor user interfaces. Travelers suffer from an overload of information and options to consider, and they lack system support in filtering information and comparing products. In this paper, we propose an innovative approach to the problem of mapping travelers’ needs to a convenient set of travel products or services. In particular, we present an on-the-move restaurant recommender system (mITR) integrated with a pre-travel planning aid system (NutKing). In this approach, firstly, the knowledge contained in a repository of past user choices is exploited to initialize the recommendation process with a set of implicit preferences. Secondly, to minimize user efforts in building a precise search query, we do not require the user to formulate a query at the beginning of the interaction, rather we involve her i n a dialogue where the traveler is encouraged to provide critiques and feedback to the system recommendations. These critiques are then incorporated by the system into a new query that tries to better model the user preferences. The approach has been validated empirically by a set of users.