Qube: Keyword Search in Peer-to-Peer Networks
Eric Yu-En Lu
We present algorithms and a new overlay topology for keyword
based search on Peer-to-Peer networks. Our approach is to develop a
simple hash function which maps all possible queries to P2P overlay
space based on Space Filling Curves. We first show that this
hash function reduces query space into an exact hypercube and is
neighbor preserving. We also show that this hash function has some
nice properties for efficient query expansion processing. We find
that the distribution of objects exhibits a form of Power Law
distribution that greatly boosts the performance of regular queries.
Finally, load-balancing extension of our scheme for both requests and
name records is proposed and preliminary results are shown.