I'm interested in any data that anyone has collected from instrumented gnutella nodes. In particular, I'd be interested if anyone has any of the following already collected:
<ul>
<li>Neighbor lifetime statistics
<li>Raw Querys
<li>Raw Query replys (esp, response file metadata, with hash info from the clients that support it)
</ul>

I'm working on some algorithms for effective lossy keyword compression and query vectorization which should allow for the implementation efficient and effective oracle nodes with specialization.

I can of course collect such data myself, but if someone hass already done it: All the easier.