Validation of Quantitative Measures for Edge Bundling by Comparing with Human Feeling

Abstract

This paper describes an analysis of the relationship between human cognition and quantitative measures for a visualization method called edge bundling.Aesthetic rules-based measures, namely, mean edge length difference (MELD), normalized MELD (NMELD), mean occupation area, and edge density distribution (EDD), for evaluating and quantifying the result of edge bundling are proposed. However, comparing these measures with human cognition has not been analyzed. Therefore, a questionnaire survey with approximately 40 respondents was conducted to clarify the relationship between human cognition and these evaluation measures. Results showed that NMELD, MELD, and EDD demonstrate robust and significant correlations with human cognition.