Efficient clustering algorithms have been discussed
in. The clustering procedure in these algorithms is based on main memory with the help of some
data structures. Since these algorithms incur fewer disk
operations, their efficiency is significantly improved.
Our objective is to study the effectiveness of the
cluster-based server selection method and’to determine
if an efficient, but not necessarily the most effective,
clustering algorithm can still improve the quality of
server selection.
As such, we adopt the widely used
K-means clustering algorithm, which is an iterative
algorithm that converges quickly. The K-means clustering algorithm starts with
K rough clusters and itera-
tively adjusts the clusters until the clusters are satisfactorily refined. Xu and Croft adopted this method and
showed that
it is quite effective, although the version
they used is
a more straightforward and faster two-pass
(iteration) algorithm.