ScatterScopes: Understanding events in real-time through spatiotemporal indication and hierarchical drilldown: VAST 2014 Mini Challenge 3 Recognition: "Honorable mention for good support for situation awareness"

The VAST Challenge 2014 MC3 featured a dataset of roughly 4000 microblog messages and 200 emergency callcenter reports produced in the fictitious city of Abila. The task was to identify relevant events and outliers in the data and highlight observations that are connected to an earlier kidnapping of employees from a company called GASTECH. In contrast to earlier challenges, the focus was on real-time data processing. The microblog messages were hosted on a web-server that simulated real-time streaming of the data during a 4.5 hour period and the challenge participants had to monitor and analyze them as they were transmitted. To tackle the challenge we developed ScatterScopes, a real-time enabled visual analytics system that fosters understanding of ongoing events by means of spatiotemporal overview and hierarchical drilldown. Trends and anomalies in space, time and content can be quickly identified with the system using interactive maps, sentiment timelines and textual search. Once interesting or suspicious subsets of elements are selected, their inherent topic structure can be further dissected based on a highly interactive treemap of message clusters. Subset selection and recombination is furthermore supported by a filter-and-flow mechanism that can also be used to formulate and test hypotheses based on Boolean logic. ScatterScopes has been used by our team to successfully identify and describe all events hidden in the MC3 data streams.