Exploring Ratings in Subjective Databases

Subjective data links people to content items and reflects who likes or dislikes what. The valuable information this data contains is virtually infinite and satisfies various information needs. Yet, as of today, dedicated tools to explore this data are lacking. In this paper, we develop a framework for Subjective Data Exploration (SDE). Our solution enables the joint exploration of items, people, and people's opinions on items, in a guided multi-step process where each step aggregates the most useful and diverse trends in the form of rating maps. Because of the large search space of possible rating maps, we leverage pruning strategies based on confidence intervals and multi-armed bandits. Our large-scale experiments with human subjects and real datasets, demonstrate the need for dedicated SDE frameworks and the effectiveness and efficiency of our approach.