Method and device for generating reference patterns from a document written in markup language and associated coding and decoding methods and devices

摘要

Reference patterns are generated which represent hierarchized data, the hierarchized data being structured in accordance with a schema and being organized in a plurality of items. Primary structural patterns are extracted from the schema, each of the primary structural patterns representing a set of structural information. A list of instances is created for at least one extracted primary structural pattern, each instance associating items of the plurality of items with a primary structural pattern. Instances of one and a same primary structural pattern are grouped into groups of instances according to a predetermined criterion applied on items of distinct instances. For each group, at least one piece of structural information common to each instance of one of the groups of instances is extracted. A reference pattern is generated from the primary structural pattern associated with the one group of instances, the reference pattern including the extracted piece of structural information.