Exploring Pros and Cons of Ranked Entities with COMPETE

Abstract

We present COMPETE, a novel approach to explore data using rankings. Utilizing rankings, which succinctly summarize the relative performance of entities, is a very intuitive way to inspecting underlying data. In this work we present an approach where users can understand the dominance of entities and interactively explore data. The approach harnesses diverse precomputed rankings, capturing many different aspects of possible user interest. For a given set of input entities, COMPETE identifies entities that are dominating or are dominated by the input, thus expanding on the relative performance of the input and giving focus on other related entities which are always better (or worse) than the input. We consider several aspects of the dominance relationship and provide different approaches that reflect these nuances. We report on the results of an experimental evaluation over data obtained from the Internet Movie Database (IMDb).