Differential Hive Plots: Seeing Networks Change

Network layouts generated by common algorithms, such as force-directed and spectral methods, are difficult to interpret because the methods are inherently unpredictable. These layouts are ineffective in comparing networks because the layouts lack perceptual uniformity—the differences in output are not proportional to differences in input. To address this, the authors introduce the differential hive plot (DHP), a layout method based on the comparison of two hive plots (HP). The DHP is a difference of two visualizations and contains nodes and edges in the difference or intersection of the two networks based on positional similarity in the input HPs.