Visual Identification and Comparison of Higher Order Properties in State Transition Sequences

Full abstract

Read the full abstract

Abstract The analysis of state transition sequences is a prevalent research topic in many domains. In this context, we introduce the term higher‐order property to describe characteristics of the analyzed data set that span beyond the local neighborhood of a single state. For the analysis of such properties, we elaborate why sequence diagrams are generally preferred over node‐link diagrams in the literature. Consequently, we provide an overview of existing adaptations to node‐link diagrams to trade‐off support for first‐order property analysis in favor of higher‐order property analysis, potentially combining the strengths of both visualization types. To better understand the impact of this tradeoff, we present a comparative study of static sequence diagrams and spline‐based node‐link diagrams for five perception tasks on small‐scale synthetic state transition sequence data. We focus on static stimuli of small‐scale data sets to model the post‐filtering perceptual process rather than an end‐to‐end analytical workflow. The study confirmed hypotheses regarding the superior user performance with the sequence diagram for the perception of higher‐order properties. To demonstrate the relevance of higher‐order property analysis for real‐world problems, we present an application scenario from the cybersecurity domain. Based on the results of our study, we apply a sequence diagram‐based prototype to this application scenario. All supplemental materials are available at https://osf.io/r4ycd/ .