A method and apparatus for generating a statistical tree representing an extensible markup language Schema (XSD) is disclosed. Components of the XSD are prioritized according to predefined rules. A root node representing the XSD is created. Pairs of child nodes are generated from the root node. Each pair comprises at least one genuine node, and each pair of generated child nodes is appended to a parent node which is a genuine node. The path to each of the child nodes from a respective parent genuine node is represented with a binary sequence. At least one genuine node is allocated to a corresponding component of the XSD, the allocation being based on the prioritization of the component. Methods, apparatus and computer program products for generating a statistical tree representing XSD, for encoding an extensible markup language (XML) document utilizing a statistical tree representing XSD, and for decoding an XML document represented by a binary encoded sequence also are disclosed.