introduction
In this project we develop a framework for analyzing and summarizing collaborative activities in a highly dynamic, context-rich social network. In everyday collaboration people need to maintain sufficient collaborative awareness in order to find opportunities for collaboration and people to collaborate with. However, this becomes difficult in a distributed, transdisciplinary collaborative situation. We seek to support collaborative awareness by extracting and representing collaborative patterns. By examining real-world collaborative events, we observe that collaborative activities are highly clustered. The clustering structures allow us to develop a compact representation of the collaborative activities that suitable for exploring temporal patterns. To extract the clustering structures, we develop a probabilistic model that factors the joint distribution of people and events into modular components. We then analyze the collaborative activities by measuring the importance of events and people within clusters based on the models. We develop a temporal representation of the clustering structures that allows users to explore patterns of their collaborative activities over time. We conduct an experiment using real-world collaborative event data. Our results demonstrate interesting observation from the complex and dynamic collaborative activities.
