This article is adapted from my 2008 honours thesis (unpublished). It presents the results of a study that investigated the use of visual analytics for aiding in the detection of a phenomenon called path preload, an anomalous pattern that can appear in message distribution on the Usenet News network.
These patterns are thought to be associated with user behaviour such as file sharing or the posting of spam to news server but are notoriously hard to detect using traditional methods of content analysis and manual assessment often relied upon by administrators to enforce news server policy and moderate content.
To overcome this challenge a visual analytic method was designed and used to examine a sample of 82,986 unique messages collected from ninety-six Usenet news groups. Message path information was used to create network visualisations that could be inspected for evidence of path preload and its potential effect on message distribution.
Results show that use of the visual analytic method can enable the successful identification of path preload and provide insights into its effect on the network that could aid in the development of techniques for automated detection and prediction of path preload.
Network analysis for identifying anomalous patterns of message distribution on Usenet News © 2008 by Harry Rolf is licensed under CC BY-ND 4.0