Abstract
Recommender systems (RS) are employed to personalize user interaction with (e.g. tourism) web-sites, supporting both navigation through large service assortments and the configuration of individual service packages. Depending on the interaction strategy, RS are either utilized to elicit users’ tastes and preferences or to stimulate desire for different offerings. In addition, as a potentially rich source of digital traces, RS also act as a repository for marketing intelligence. Web-usage mining is an accepted approach to analyse web-usage behaviour based on information traces left by the web-user (Mobasher, 2007). This paper proposes an empirically tested approach which combines typical web-log data with user feedback gathered by an interactive travel advisory system developed for an Austrian spa-resort. The proposed approach focuses on evaluating the RS with respect to efficiency, effectiveness and actionable marketing intelligence.