Análise lexicográfica da produção acadêmica da Fiocruz: uma proposta de metodologia

• Autor(a) principal: Lima, Jefferson da Costa
• Data de Publicação: 2016
• Tipo de documento: Dissertação
• Idioma: por
• Título da fonte: Repositório Institucional do FGV (FGV Repositório Digital)
• Texto Completo: http://hdl.handle.net/10438/17458
• topic: Processamento de linguagem natural; Mineração de textos; Descritores em ciências da saúde; Matemática; Processamento da linguagem natural (Computação); Mineração de dados (Computação)

With the objective to meet the health needs of the population, a huge amount of publications are generated each year. Scientific papers, thesis and dissertations become available digitally, but make them accessible to the user requires an understanding of the indexing process, which is usually made manually. This work proposes an experiment on the feasibility of automatically identify valid descriptors for the documents in the field of health. Are extracted n-grams of the texts and, after comparison with terms of vocabulary Health Sciences Descriptors (DeCS), are identified those who can act as descriptors for the works. We believe that this process can be applied to classification of document sets with deficiencies in their indexing and, even, in supporting the re-indexing processes, improving the precision and recall of the searches, and the possibility of establishing metrics of relevance.