Improved the clustering quality of the massive extensible markup language (XML) document set, a vector matrix iterative self-organizing assistant clustering algorithm of XML documents (VMISACAX) is proposed. Bases on the XML key, the algorithm transforms the XML document into a vector matrix and carries out the optimizations of canceling, dissociating and uniting etc. In order to improve the convergence of the algorithm, a assistant strategy is imported to shorten the algorithm time under settling for clustering, to obtain the best result of clustering by XML key's weights, but it doesn't alwa...