Lexical Analysis Using Regular Expressions for Information Retrieval from a Legal Corpus

Abstract

This article presents part of the work carried out in the framework of a research that aims to optimize an Information Retrieval System, by means of its specialization for the retrieval of legal documents. One of the fundamental sub-processes in this type of system is lexical analysis, in which indexing techniques are applied. These techniques involve extracting a series of concepts representative of the topics covered in a document, and then using them as access points for retrieval. This article describes a proposal for the extraction of information and identification of dates and references to named entities, such as File No., Resolution No., Article No. of Law XXX, which refer to the legal norm in force and are widely used in different judicial documents. For the recognition of such named entities, the process employed the definition of patterns using Regular Expressions, a way of representing a language in a synthetic form, applying a set of rules. From this, the terms obtained are stored in a matrix of terms/documents. This paper also describes the algorithms used during the validation of the proposed solution and presents the experimental results that show that by applying this method a significant reduction in the size of the inputs to the matrix can be achieved.