Service-Oriented Architecture for automatic markup of documents. An use case for legal documents

The problem of information extraction and automatic markup of plain text to XML, has been resolved partially in a specific domain of legal documents. Techniques such as named entity recognition, hierarchy detection of text sections and others has led to partially identify and retrieve different kind of information inside non structured documents. In this paper we introduce different interconnected components, the NLP techniques used on each component and the workflow needed for processing a plain text document and to generate a new full marked XML version of the document. The generated XML complies with the schema legal standard Akoma-Ntoso and is highly enriched with named entities, semantic URIS, structural sections, lists and elements sequences, between others. As an use case we analyze the experience of the Library of Congress of Chile in the context of the 'History of Law project' and Parliamentary Labor, where these architecture had a key role in order to accomplish the final product and results of processing and marking up different types or models of documents used in the legislative process.