IBM Israel Research Seminars

My talk is based on one part in my Ph.D. dissertation “Gatekeepers and Gatekeeping Mechanisms in Networks”. The project originally identifies and classifies major social, technical, economic, business, and legal gatekeeping mechanisms and gatekeepers in the Internet. The study employs a theoretical framework to conceptualize gatekeeping and exhibits an empirical investigation of a large dataset. Through using content analysis and data mining it unveils and explicates the various paths of controlling information and the methods and motivations at work in exercising this control.

On the one hand, my theoretical typology exposes a rather firm tendency of centralization of information control by a small number of enterprises in the Internet. Gatekeeping through this prism is conceived as discouraging the free flow of information on the net. On the other hand, the empirical part presents a great need by virtual communities to set various types of gatekeeping mechanisms that enable them to conduct a virtual flow of information and prevent sabotaging the operational activities on their platforms. It follows that the Internet business models must provide solutions for the management and governance of the information flow. These solutions should balance the need for a largely unsupervised and free mobility of users within the net with a secured and efficient ability of users to conduct their activities online.

More specifically, I used data mining to examine 717 virtual communities while relying on a massive database of 2 million observations that include information traffic as well as forum and user characteristics within these communities over a period of three years. My dataset explores messages that were blocked by managers, their content, reasons for blocking, users who were removed by the service enablers of these communities, and the characteristics of messages that were unblocked. The research examines explanatory variables that can explain information control exercised by virtual community enablers, and provides a predictive model that can flag abnormal online consumer behavior.