TY  - CHAP
TI  - Visualizing Life in a Graph Stream
AU  - Abello, James
AU  - DeSimone, David
AU  - Hadlak, Steffen
AU  - Schulz, Hans-Jörg
AU  - Sumida, Mika
T2  - Big Data of Complex Networks
A2  - Dehmer, Matthias
A2  - Emmert-Streib, Frank
A2  - Pickl, Stefan
A2  - Holzinger, Andreas
AB  - We introduce a simple and useful view for observing graph streams. They are viewed as collections of edge events where each edge has associated a set of time-dependent statistics that include firing rate, recency, and persistence. The activity rate of any subgraph is expressed as an aggregation of its corresponding edge statistics. Salient subgraphs are detected by isolating through time those edges whose activity rate deviates substantially from the activity rate of the entire stream. These salient subgraphs exhibit some peculiar "herding" and "straying" behaviors that are humanly interpretable. The vertices involved in the creation of these salient behaviors cover a substantial portion of the entire graph stream. This coverage can be subject to both human and computer verification. All our computations are incremental and are accompanied by a visualization platform that integrates dynamic node link views of "recent" graph substreams with a tape view of the Top-K edge statistics to provide a compact overview of the graph stream. This platform has also been coupled to our modular Degree-of-Interest system for a closer investigation of those patterns found in the overview. We use Twitter data to illustrate our tools, but our approach is by no means confined to microblog data.
DA  - 2016///
PY  - 2016
SP  - 293
EP  - 312
PB  - Chapman and Hall/CRC
SN  - 978-1-4987-2361-9
UR  - https://www.taylorfrancis.com/books/e/9781498723626/chapters/10.1201%2F9781315370736-19
ER  -