Statistics of real published xml data

Hello everyone,

    I want to do some investigation on the statistics of real  
published xml data, especially for data in xml-based publish/subscribe  
system. Please give me some suggestions if you have this information.  
Thank you.

Yang