Study on techniques of clustering XML data

Analyzing the characteristics of data access in DOM standards, two methods are proposed for clustering XML data, which are based on the parent-child and sibling relationship separately. Then the probabilities of I/O occurrence of the two clustering methods are analyzed with respect to two typical operations of traversing over a tree, and two corresponding path expression algorithms to traverse over DOM tree are given, namely depth-first and breadth-first algorithms. A performance evaluation based on XMark and XMach is carried out as testing references.