Progressive Disclosure of User Preferences in a Tourism Recommender System

We present a web-based travel recommender system that gives users direct control over the accuracy-privacy trade-off by progressively eliciting preferences across nine optional steps and showing value after each. Users can stop at any time, receive ranked destination suggestions, and decide whether to disclose more information to refine results, which links disclosure to visible utility and preserves agency. The adaptive hybrid engine of the system combines content-based filtering with item-item collaborative filtering that activates once ratings are available, ensuring meaningful recommendations with partial input and improving as profiles deepen. An empirical user study shows a strong positive relationship between the amount of information disclosed and recommendation precision, while maintaining high utility overall: 93.33% of recommendation sets contained at least one liked item. The contributions are a progressive onboarding interface, an adaptive hybrid recommendation approach that remains effective under sparse inputs, and empirical validation that user-centric disclosure can manage accuracy and privacy transparently, fostering trust and user agency