Online User Activities Discovery Based on Time Dependent Data

Network evolution is a hot research topic especially when social networking has become an important Web application. The access histories of Web users which contain the users traces' on a social network have not been considered useful data. However, they may reveal more about the network's connectedness if the history's time-sensitive characteristic is analyzed and studied. In this paper, we model the user's daily activities in a time series model to reflect the dynamic nature of a social network due to various user behavior patterns over a period of time. We begin to study the activity pattern for a single user. We then expand that study over the whole network. Through the model, we can quantitatively analyze the user's contribution to the social network and predict the user's response when there is a new action by another user.