Visualization of large temporal social network datasets

Official URL: http://192.168.1.20/record=b1225704 (Table of Contents)

Social network datasets consist of what sociologists call ‘social structures’, accumulation of all communication channels that social actors share ideas and information between each other. Social network analysis reveals characteristics and properties of social networks by applying specific metrics. Although, size of a real-life social network dataset can reach millions of relations belong to millions of social actors with large temporal dimension, existing information visualization tools can represent at most several thousands of these actors. This thesis presents a conceptual design study focused on visualization of large temporal social network datasets with a novel visualization method. Proposed technique combines Ideal Gas Law (IGL) with Jacob Moreno’s theory of The Cannon of Creativity to layout social network datasets in 3D hyperbolic space and can render 50,000 social actors at interactive speed. A proof-of-the-concept program is developed around this technique allowing users to perform several analysis tasks on temporal social network datasets. Users can explore the network, control the amount of visual clutter, and identify communication anomalies in run time. Moreover, they can search a specific actor and visually follow her communication pattern. The effectiveness of proposed technique is presented with case and usability studies performed using generated and real-life datasets. In particular the Enron email dataset (323,073 emails, 19,898 email addresses over four years) and 20 Newsgroups (44,797 postings, 20 news groups and 5417 email addresses over one month) datasets are analyzed.