Indexação automática de acórdãos por meio de processamento de linguagem natural

AUTOR(ES)
DATA DE PUBLICAÇÃO

2007

RESUMO

Indexing is an area that has been researched for a long time, crucial tool to information retrieval quality. Automatic indexing is a more recent area, in which results are improved with the evolution of artificial intelligence techniques. As judicial claims increase repeatedly, Countrys courthouses, Tribunal de Justiça do Distrito Federal e dos Territórios in particular, have to search for mechanisms, in their majority based on information technology, to speed up their workflows. Automatic indexing offers strategic gains to courthouses, diminishing their personnel costs, improving research quality and accelerating their organizational processes. This research inserts itself in this area and proposes a sentence automatic indexing model which fulfills these needs. NLP utilization is chosen instead of strictly statistic methods with the purpose of achieving better results. In order to do so, tools have been developed which build a corpus of portuguese language containing common juridical vocabulary and automatically index a criminal law jurisprudence base from TJDFT. Results demonstrate that automatic indexing using NLP obtains equivalent revocation and precision levels than manual indexing. Between this research contributions are mentioned the developed tools, the automatic indexing methodology proposed and the study of a very little researched area in portuguese language.

ASSUNTO(S)

indexação automática ciencia da informacao jurisprudence jurisprudência indexação manual manual indexing processamento de linguagem natural natural language processing automatic indexing information retrieval recuperação da informação

Documentos Relacionados