A Prediction Model of Privacy Control for Online Social Networking Users

With the growing popularity of social network sites (SNS), organizations have started to leverage them for encouraging both personal and professional data sharing. However, inherent privacy problems in social networks have become a concern for organizations deploying them. So companies have started investing in systems for evaluating employees’ behaviors on SNSs. In evaluating employees’ behaviors on SNSs, this study aims at developing a mechanism for learning users’ behaviors on SNS and predicting their control of privacy on SNS. Privacy prediction is based on the revelation of actual privacy characteristics of users through the analysis of their SNS usage patterns. Using the Design Science research methodology, this study presents the design and instantiation of a prediction model that is trained using survey data and SNS data of graduate students from a prominent Northeastern University in the United States, which is used to generate class labels associated with their privacy control. The prediction model provides a data analytics component for reliable predictions of users’ privacy control using Machine Learning algorithm SVM and a randomized ensemble of decision trees. The results suggest that the prediction model represents a reliable method for predicting privacy control based on user actions on SNS.