Using Linguistic Information and Machine Learning Techniques to Identify Entities from Juridical Documents

Abstract

Information extraction from legal documents is an important and open problem. A mixed approach, using linguistic information and machine learning techniques, is described in this paper. In this approach, top-level legal concepts are identified and used for document classifica- tion using Support Vector Machines. Named entities, such as, locations, organizations, dates, and document references, are identified using se- mantic information from the output of a natural language parser. This information, legal concepts and named entities, may be used to popu- late a simple ontology, allowing the enrichment of documents and the creation of high-level legal information retrieval systems. The proposed methodology was applied to a corpus of legal documents - from the EUR-Lex site – and it was evaluated. The obtained results were quite good and indicate this may be a promising approach to the legal information extraction problem.