... (Pantel and Lin, 2002;Sch¨utze, 1998), there are other related efforts onword sense discrimination (Dorow and Widdows,2003; Fukumoto and Suzuki, 1999; Pedersen and Bruce, 1997).In (Pedersen and ... selected feature subset is impor-tant and complete, cluster structure estimated fromdata subset in this feature space should be stable and robust against random sampling. After deter-mination ... estimate clusternumber and achieve better performance compared with CGD algorithm. This paper is organized as follows. In section2 we will introduce our word sense learning al-gorithm, which...