Visual Analysis of Dynamic Networks with Geological Clustering Adel Ahmed∗ Xiaoyan Fu† Seok-Hee Hong‡ School of Information Technologies University of Sydney, Australia National ICT Australia, Australia National ICT Australia, Australia School of Information Technologies University of Sydney, Australia National ICT Australia, Australia Quan Hoang NguyenĐ Kai Xuả School of Computer Sciences and Engineering University of NSW, Australia National ICT Australia, Australia A BSTRACT Many dynamic networks have associated geological information Here we present two complementing visual analysis methods for such networks The first one provides an overview with summerized information while the second one presents a more detailed view The geological information is encoded in the network layout, which is designed to help maintain user’s mental map We also combined visualization with social network analysis to facilitate knowledge discovery, especially to understand network changes in the context overall evolution Both methods are applied to the “History of the FIFA World Cup Competition” data set Keywords: Network Visualization, Visual Analytics, Dynamic Network, Temporal Network, Hierarchy, Clustering, Centrality Index Terms: H.5.2 [INFORMATION INTERFACES AND PRESENTATION]: User Interfaces—Theory and methods; I.3.6 [Computer Graphics]: Methodology and Techniques—Interaction Techniques I NTRODUCTION Many dynamic networks have geological information One example is the email communication networks between people, in which emails can be sent from different locations such as home or office Another example is the world trading network where nodes are countries and edges are the trading between them Inherently each country has its geological location It is important to consider the geological information when analyzing such dynamic networks There are several existing methods for visualization of dynamic networks [1, 3], but none of them considers the geological information A recent relevant work by Shneiderman and Aris [6] addresses the geographical information in network visualization, but it is for static networks Also, there are several work on combining network visualization with social network analysis [4, 5], but they not consider dynamic networks and geological information Here we propose two visual analysis methods for dynamic networks with geological information by combining visualization with social network analysis These two methods complement each other: The first one presents the overview of a dynamic network by showing only the summerized information, and the second one includes the details of every network change In both methods, nodes are clustered according to their geological location (i.e., the graph ∗ e-mail: adel.ahmed@nicta.com.au xiaoyan.fu@nicta.com.au e-mail: shhong@it.usyd.edu.au Đ e-mail: quanhn@cse.unsw.edu.au ả e-mail: kai.xu@nicta.com.au † e-mail: layout considers the geological information), and the results of social network analysis are mapped visually to facilitate its analysis, especially for network changes in the context of overall evolution The two methods are applied to a real-world data set: the History of the FIFA World Cup Competition T HE DATA S ET AND S OCIAL N ETWORK A NALYSIS The FIFA World Cup Competition History data set contains the results of all the matches played in the final rounds since its founding in 1930 The World Cup is organized every four years, but due to the World War II, only 18 tournaments have been held so far There are in total 79 countries that have ever joined the final rounds, and can be clustered based on their geographic locations into six football federations: AFC (Asia), CAF (Africa), CONCACAF (North America), CONMEBOL (South America), OFC (Oceania), and UEFA (Europe) The data set can be represented as a dynamic network that consists of a series of directed graphs (one for each world cup) whose nodes are countries and edges are the matches with winning team pointing to the losing one It is easy to see that: First, the network is dynamic, as the nodes and edges of a world cup graph changes from one year to another; second, the network is temporal, as each world cup graph has a time stamp, and thus their ordering is fixed by the time series; finally, the network has a geological clustering structure according to country location or football federation membership Centrality index is an important concept in social network analysis for analyzing the importance of actors embedded in a social network [2, 7] In our methods, we used centrality analysis to show the strong performers over the years and their performance change over the time We computed the centralities for each world cup and the whole data set First, we construct a series of directed graph Gi , i = 1, , 18, one for each World Cup Then, we construct a union graph G = G1 ∪ G2 ∪ ∪ G18 for analyzing the global performance Among the many centrality measurements available, we chose to include the results of degree centrality—defined by the number of edges incident to a node—because it is a clear indicator of team performance Note that our methods can be easily applied to other centralities W HEEL L AYOUT In this first method, we place each country that ever joined world cup in the outermost circle of a wheel and represent each world cup as a concentric inner circle The radius increases with the World Cup year: The circle corresponding to the first ever world cup (year 1930) is the inner most circle near the center, and the circle corresponding to the latest world cup (year 2006) is placed just inside the outermost circle The performance of a country at a specific year is shown as the size of the node located at the intersection of the inner circle for that year and the radius pointed to the country node in the outermost circle The node size is decided by its degree centrality value and nodes on the same World Cup circle have the same IEEE Symposium on Visual Analytics Science and Technology 2007 October 30 - November 1, Sacramento, CA, USA 978-1-4244-1659-2/07/$25.00 ©2007 IEEE 221 color To show the geological clustering, the wheel is divided into wedges based on the football federations, and then each country is placed in its federation wedge Figure shows the result of applying this method to the world cup history data set An animation is included in the accompanied video to illustrate the evolution of team performance Figure 2: Radial layout Figure 1: Wheel layout at both the embedding position and the node size For example, in year 2002 (Figure 3) South Korea performed unexpectedly well: it has one of the largest node size, although it is placed in the middle circle For team performance evolution, we can see from the video that Brazil actually did not perform very well in the early years of World Cup history, although it is undoubtedly the best performer now It can been seen clearly from the wheel layout the evolution of country performance over the years For example, it is clear that Brazil, Germany and Italy are among the strongest teams in the world cup history, as they have many large circles along their radius Moreover, one can compare the performance between countries of a specific year, by inspecting the node size along the inner circle representing that world cup Similarly, one can easily compare the performance between the continents and inside each continent For example, in general the European countries performed better than other continents Among the Asian countries, South Korea performed relatively well The wheel layout only presents an overview and hide the information such the network topology A radial layout is designed to complement it R ADIAL L AYOUT This layout displays the network topology and team performance at every world cup We followed the radial drawing convention from the social network analysis for displaying centrality: We place the node with the highest centrality value at the center, and then place the nodes with the next highest centrality values using concentric circles However, we made the following important modifications First, instead of using the exact centrality value to define concentric circle—which may end up with too many circles—we discretized all the countries into four groups: one overall winner plus groups (i.e strong, medium and weak) based on their centrality values Then we place the strong group in the innermost circle, medium group in the middle circle, and the weak group in the outermost circle Second, to help maintain user’s mental map, we layout the union graph first, then use its node location to layout the graph of every world cup, so every country appears at the same location in different world cup graphs Finally, the nodes in the union graph layout are partitioned according to their federation regions to show the geological clustering Similar to the wheel layout method, the size of each node is based on its degree centrality The graph of each world cup is shown on a 2D plane, and these planes are stacked together in the third dimension to show the entire history (Figure 2) The accompanied video provides views of the radial layout from different angles In the radial layout, we can analyze each team’s performance of a specific year in the context of its overall performance by looking The video can be downloaded from http://www.cs.usyd.edu au/˜visual/kaixu/worldcup_video.zip 222 Figure 3: World Cup 2002 R EFERENCES [1] U Brandes and S R Corman Visual unrolling of network evolution and the analysis of dynamic discourse Information Visualization, 2(1):40–50, 2003 [2] U Brandes and T Erlebach Network Analysis: methodological foundations Springer, 2005 [3] U Brandes, M Hoefer, and C Pich Affiliation dynamics with an application to movie-actor biographies In Proc Eurographics/IEEE-VGTC Symp Visualization (EuroVis ’06), pages 179–186, 2006 [4] C Chen The centrality of pivotal points in the evolution of scientific networks In IUI ’05: Proceedings of the 10th international conference on Intelligent user interfaces, pages 98–105, New York, NY, USA, 2005 ACM Press [5] A Perer and B Shneiderman Balancing systematic and flexible exploration of social networks IEEE Transactions on Visualization and Computer Graphics, 12(5):693–700, 2006 [6] B Shneiderman and A Aris Network visualization by semantic substrates IEEE Transactions on Visualization and Computer Graphics, 12(5):733–740, 2006 [7] S Wasserman and K Faust Social Network Analysis: Methods and Applicaitons Cambridge University Press, 40 West 20th Street, New York, NY 10011-4211, USA, 1st edition, 1995 ... au/? ?visual/ kaixu/worldcup_video.zip 222 Figure 3: World Cup 2002 R EFERENCES [1] U Brandes and S R Corman Visual unrolling of network evolution and the analysis of dynamic discourse Information Visualization,... Eurographics/IEEE-VGTC Symp Visualization (EuroVis ’06), pages 179–186, 2006 [4] C Chen The centrality of pivotal points in the evolution of scientific networks In IUI ’05: Proceedings of the 10th international... their federation regions to show the geological clustering Similar to the wheel layout method, the size of each node is based on its degree centrality The graph of each world cup is shown on a 2D