What we want to do is stated by Staging and Evaluating Public Performances as an Approach to CVE Research by Steve Benford, Mike Fraser, Gail Reynard, Boriana Koleva and Adam Drozd (The Mixed Reality Laboratory):

System instrumentation and analysis of logs developers instrument the underlying CVE platform to log (and timestamp) as much data as possible at participants machines and as it passes across the network. This data might include all avatar movements, interactions with virtual objects, audio packets and text messages. These logs can then be analysed statistically in order to uncover significant patterns of user activity (e.g., frequencies, distributions and correlation of movements and communication). The results can support or contradict other observations (participants own reflections and ethnography), can inform models of system and network performance, and can suggest new technical designs and optimisations.

They also say: "of system logs is also more problematic than it need be. At present, there is no agreed format for log data and no readily available suites of analysis tools.", that's something I want to work on: defining way to analysz collaborative mobile activities.