The algorithm above presented is the bisecting version of the general K-means
algorithm. This bisecting algorithm has been recently discussed and emphasized in [17]
and [19]. In these works it is claimed to be very effective in document-processing
problems. It is here worth noting that the algorithm above recalled is the very classical
and basic version of K-means, also known (see [10, 12]) as Forgy’s algorithm (with a
slight modification of the initialization step). Many variations of this basic version of the
algorithm have been proposed, aiming to reduce the computational demand, at the price
of (hopefully little) sub-optimality. Since the goal of this paper is to analyze convergence
properties and clustering performance, this original version of the K-means algorithm is
the most interesting and meaningful.