Published in Signal Processing Magazine, IEEE volume 30, issue 6, 157-162.
Community safety as a social issue has expanded its reach to web forums, portals and dedicated sites. This paper presents our study of 230 community safety groups whose members communicate through the Nation of Neighbors website. We analyze the patterns of activities within these communities along with their temporal dynamics. We demonstrate both feature-based and temporal analyses of the communities aiming at discovering the characteristics that make such communities successful. We use ManyNetsís capability to visualize the overview of multiple networks at once, demonstrating the value of visual analytics for community managers to better understand their communities. Using previously-developed health metrics we distinguish the successful communities, observe the influence of leaders in those communities and establish that larger communities are reporting more crime incidents rather than having discussion on other topics. To our surprise, we did not observe any strong association between the involvement of Law Enforcement personnel and activeness of the communities.