Besides crawling, people can also study Online Social Networks by monitoring the corresponding network traffic. Benevenuto et al. [6] analyze the user behavior of Online Social Networks based on detailed clickstream data obtained from a social network aggregator, as illustrated in Fig. 1. In [6], the clickstream data was collected over 12 days with HTTP sessions of 37,024 users who accessed popular social networks. This article defines and analyzes the Online Social Networks session characteristics:
• The frequency of accessing Online Social Networks
• Total time spent on Online Social Networks
• Session duration of Online Social Networks
Through the clickstream data, user activities are also identified. Forty-one types of user activities are classified into nine groups, and the popularity of different activities and the traffic bytes are analyzed. Interestingly, it is found that silent or latent interactions such as browsing account for more than 90 percent of user activities. Also, they show how users have different activities in different Online Social Networks. They also characterize how users transit from one activity to another using a first-order Markov chain.