A visualization approach to performing task analysis of time series event log data from a microworld simulation

Abstract

This paper presents a novel visualization technique to perform a task analysis using only information from data logs collected from users interacting with the system. Traditional task analysis includes data acquired from system specifications, observations of the user interacting with the system, and interviews with the user. There are however situations, in which this information is not available such as forensic investigations. In situations in which these other data sources are available, the visualization approach provides an additional tool the analysist can use to further their understanding of the system and begin qualitatively and quantitatively describing a user’s behavior. As a demonstration of this technique, data logs from a microworld simulator, the Grid Game, were analyzed using a custom designed python based graphing tool. The Grid Game is a gamified and simplified simulation of an electric microgrid. Teams of players compete by maintaining and optimizing their grid configuration in the presence of cyber-attacks. Several different visualization graphing techniques were examined to identify their suitability for representing the grid event data in a format capable to support an analyst performing a task analysis. The most suitable graphing technique for this particular use case was a discrete event graphing technique including a common time dimension to provide performance based context to the graphical depiction of logged simulation events.