Desire: A Dynamic Approach for Exploratory Search Results Recommendation

Abstract

Personalized search aims to capture different users' needs in order to provide them with relevant results considering their individual interests. Therefore, personalized search systems store information about users' preferences and interests to create individualized profiles that can be integrated into the retrieval process. Several approaches have been proposed to dynamically capture the real user interest and achieve such objective. However, works so far have restricted themselves to the results page, failing to explore the possibility of recommending search results while the user navigates through the search space. Thus, we propose DESiRe, a dynamic approach for search results recommendation that is able to present unseen relevant results while the user browses the retrieved search space. Evaluation results showed an overall improvement of 88% for the ranking quality when compared to Google. The recommendation process was equally effective, providing high quality recommendations with a relatively large number of unseen results.