Mining and Visualizing the Evolution of Subgroups in

Mining and Visualizing the Evolution of Subgroups in Social Networks Falkowsky, T. , Bartelheimer, J. & Spiliopoulou, M. (2006) IEEE/WIC/ACM International Conference on Web Intelligence, pp. 52 -58 Presented by Danielle Lee

Outline n n n Problem Research Purpose Data Set First Approach : Statistical Analyses and Visualization for relatively stable communities Second Approach : Detection of the subgroup evolution in high fluctuating communities

Problem n n n A community has rather stable structure with a small amount of fluctuating members and they participate in over a long time. Another community has high dynamic structure whose members and their networks keep changing over time. Different community detection and visualization methods are needed.

Research Purpose n To propose statistical method and visualization to analyze the formation of subgroups and the timely change of online communities on the level of sub-groups

Data Set n n n Taken from an online international student community in the University of Magdeburg. About 1000 members from more than 50 countries 250, 000 guestbook entries over a period of 18 months

Evolution of Subgroups in Static Structure (Contd. ) n Mining for subgroups in Social Networks 1) 2) 3) Partitioning data by time axis Weight graph Gt of interactions between individuals for each time windows is built. Hierarchical edge betweenness clustering of the graph is applied in each time window

Evolution of Subgroups in Static Structure (Contd. ) Subgroups Communication within one community Detailed information at a certain time point time

Evolution of Subgroups in Static Structure (Contd. ) n Analyzing Subgroup Dynamics ¡ Track a detected subgroup over time by measuring the structural equivalence n n n ¡ Stability Density and cohesion Euclidean distance Correlation coefficient Group activity The measures are computed for each time window n n Fixed : A chosen time window is compared with all other windows Periodical : Each time window is compared to the previous time window

Evolution of Subgroups in Static Structure Kinds of Measurement Each Subgroup

Dynamics of Communities with Fluctuating Members (contd. ) n Clustering subgroups as a community ¡ ¡ ¡ Establish a graph of subgroups to denote similarity about them Similarity have been discovered as the overlap of members between two subgroups Two subgroups are similar if their overlap exceeds a given threshold.