Visual Analytic Decision-Making Environments for Large-Scale Time-Evolving Graphs

Abstract

Abstract Decision makers in multiple domains are increasingly looking for ways to improve the understanding of real-world phenomena through data collected from Internet devices, including low-cost sensors, smart phones, and online activity. Examples include detecting environmental changes, understanding the impacts of adverse manmade and natural disasters, and obtaining marketing intelligence on company profiles. Real-world observations from streaming data sources such as sensors, click streams, and social media are becoming increasingly common in many domains, including cybersecurity, infectious disease surveillance, social community networks, and web-based recommendation systems. An emerging approach is to leverage graph-based modeling of these events, to understand ongoing trends and predict future ones. Data scientists are faced with the challenge of analyzing large-scale graphs that are changing dynamically, while existing tools and metaphors for data collection, processing, storage, and analysis are not suitable for handling large-scale evolutionary graphs. This chapter describes visual analytics as a cognitive computing approach to improve decision making with large-scale dynamic graphs. We provide a conceptual introduction to time-varying graphs and various components of the visual analytics that affect the performance of decision support systems, including data management, analytics, visualization, and visual interaction. We provide techniques to improve the performance of each of these components in an integrated visual analytics framework. We also describe a visual graph analytics sandbox architecture and sample applications implemented within it.