An Effective Distributed k-Means Clustering Algorithm Based on the Pretreatment of Vectors' Inner-Product