Visualizing Concordance of Sets

When there is more than one perspective to interpret a dataset, concordance (or discordance) between the result sets from the different perspectives plays an important role in getting refined results. For example, several clustering algorithms generate different results for the same input. To get useful insights, users need to combine different perspectives by checking concordance between those results. In this paper, we present an interactive visualization tool called ConSet, where users can effectively examine multiple sets at once. It uses permutation matrix visualization to enable users to easily identify similar sets. In addition to a standard Venn diagram, we introduce a Fairy diagram that allows users to compare two or three sets without inconsistencies. We conducted a qualitative user study to evaluate how our tool works in comparison with a traditional set visualization tool based on a Venn diagram. We found that users performed better with ConSet than with the traditional interface for many tasks and most users preferred ConSet. CR