Evaluating overall quality of graph visualizations based on aesthetics aggregation

Abstract

Aesthetics are often used to measure the layout quality of graph drawings and it is commonly accepted that drawings with good layout are effective in conveying the embedded data information to end users. However, existing aesthetic criteria are useful only in judging the extents to which a drawing conforms to specific drawing rules. They have limitations in evaluating overall quality. Currently graph visualizations are mainly evaluated based on personal judgments and user studies for their overall quality. Personal judgments are not reliable while user studies can be costly to run. Therefore, there is a need for a direct measure of overall quality. In an attempt to meet this need, we propose a measurement that measures overall quality based on individual aesthetics and gives a single numerical score. We present a user study that validates this measure by demonstrating its sensibility in detecting quality changes and its capacity in predicting the performance of human graph comprehension. The implications of our proposed measure for future research are discussed.