Temporal Visualization of Social Network Dynamics: Prototypes for Nation of NeighborsIntroductionRelated StudiesSocial Network Dynamics Visualization – IdeasSocial Network Dynamic Visualization SystemsSpreadsheet-Based Approach Using NodeXLTempoVis: Adding Time-Based Interactive ExplorationDiscussions and ConclusionsReferencesTemporal Visualization of Social NetworkDynamics: Prototypes for Nation of NeighborsJae-wook Ahn1, 2, Meirav Taieb-Maimon2, 3, Awalin Sopan1, 2,Catherine Plaisant2, and Ben Shneiderman1, 21Department of Computer Science2Human-Computer Interaction Lab, University of Maryland,College Park, MD, USA{jahn,awalin,plaisant,ben}@cs.umd.edu3Ben-Gurion University of the Negev, Beer-Sheva, [email protected]. Information visualization is a powerful tool for analyzing thedynamic nature of social communities. Using Nation of Neighbors com-munity network as a testbed, we propose five principles of implement-ing temporal visualizations for social networks and present two researchprototypes: NodeXL and TempoVis. Three different states are defined inorder to visualize the temporal changes of social networks. We designedthe prototypes to show the benefits of the proposed ideas by letting usersinteractively explore temporal changes of social networks.Keywords: Social Network Analysis, Information Visualization, SocialDynamics, User Interface, Temporal Evolution.1 IntroductionInformation visualization is a powerful tool for analyzing complex social networksso as to provide users with actionable insights even against large amount of data.Numerous social network analysis (SNA) methods have been coupled with visu-alization to uncover influential actors, find helpful bridging people, and identifydestructive spammers. There are many research tools and a growing numberof commercial software packages, some designed for innovative large scale dataanalysis, while others deal with common business intelligence needs. However,few approaches or tools sufficiently address the problem of how to analyze thesocial dynamics of change over time visually. Communities are not static. Likeliving organisms, they evolve because of cultural, environmental, economic, orpolitical trends, external interventions, or unexpected events [4]. Technologicaldevelopments have also had strong impacts on social changes, a phenomenonthat has become influential with the arrival of mobile communications devicesand social networking services.Nation of Neighbors (http://www.nationofneighbors.com) (NoN) is a newweb-based community network that enables neighbors to share local crime, sus-picious activity, and other community concerns in real time. The NoN developersJ. Salerno et al. (Eds.): SBP 2011, LNCS 6589, pp. 309–316, 2011.c Springer-Verlag Berlin Heidelberg 2011310 J.-w. Ahn et al.have achieved an admirable success with “Watch Jefferson County” that empow-ers community members to maintain the security in their local neighborhoods.It began in Jefferson County, WV, but the NoN team is expanding their effortsacross the U.S. in many communities. We are collaborating with them to provideappropriate tools that can help community managers explore and analyze thesocial dynamics embedded in their social networks.This paper introduces our first efforts to visualize the temporal social dynam-ics on top of the NoN testbed. Due to the nature of the social dynamics, weneed a method to interactively compare the time slices of social networks. Inprevious attempts, we combined statistics and the visualization [12] for interac-tive network analysis and to support multiple network comparisons using tabularrepresentations [3]. The current work integrates the best of these two approachesby using the graphical representations and statistics to enable researchers andcommunity managers to compare networks over time.2 Related StudiesThe importance of capturing the network change over time has been the origin ofprevious attempts of temporal network visualizations. Two common approacheswere used. The first approach plots network summary statistics as line graphsover time (e.g. [1,7]). Due to its advantage in detecting the increase and de-crease of certain statistics, many systems adopted this idea. In the VAST 2008mini challenge 3:“Cell Phone Calls” [8] various teams (e.g. SocialDynamicsVisand Prajna teams) used similar approaches such as line graphs to characterizechanges of the cell phone call social network over ten days. SocialAction in-tegrated a time-based interactive stacked graph to display a network propertysuch as degree. Each stack represented a node and the thickness of each stack ona given day represented its degree. The interactive visualization facilitated theexploration of the change in the degree of each node over time.The second approach is to examine separate images of the network at eachpoint in time. Powell et al. [13] used such a snapshot approach to show networkchanges over time. They distinguished new arrivals with triangles and incumbentwith circles. They also used size encoding for degree; a shortcoming of such sizeand shape encoding might be that the new arrivals having small degree are verysmall and sometimes the triangles and circles are too small to be distinguish-able. Furthermore, in snapshot view we cannot compare a network easily in twotimeslots as the layout algorithms can dynamically change the node positions.Durant et al. [2] presented a method for assessing responses and identifying theevolution stages of an online discussion board. Their snapshot-based networkvisualizations could show different node positions between time.Recent work offered improvements in static visualizations by applying dy-namic visualization such as network “movies.” Moody [11] distinguished flip-book style movies where node-positions were fixed but connectivity formationwas captured and dynamic movies where node-positions changed over time. Con-dor (or TeCFlow) [5,6] introduced sliding time frames and animated the networkTemporal Visualization of Social Network Dynamics 311changes inside a specific time frame with/without the history before that. How-ever, approaches which use animations might be distracting for users to trackchanges in the network. Specifically, it might be difficult to track new nodeswhen the position of the nodes and edges keep changing. Human eyes can missminute but important changes in visual spaces. For example, users can only no-tice the overall growth