Monitoração de procedimentos em um tesauro multilíngüe

AUTOR(ES)
DATA DE PUBLICAÇÃO

2006

RESUMO

Abstract: To build a thesaurus with synonyms-based classes and include acceptions in medical domain is not a trivial task, due to the inherent terminology complexity. In this dissertation we show some highlighted problems present in the manual mode thesaurus construction and describe an approach to help the lexicographers in its construction and maintenance process. We introduce the conception of thesaurus management anomalies as sequence of such actions done by the thesaurus curators that consume effort without any positive impact on the quality of the thesaurus. In this research 4 kinds of frequently incorrect procedures were mapped on the thesaurus: relationship, type, delimitation and permanence anomalies. The main goal of this research is the identification of inherit problems in the process of thesaurus construction and maintenance due to the lack of lexicographers knowledge as well as the difficulty to solve linguistics phenomena. As workbench we used the last versions of MorphoSaurus Project s thesaurus representing 86 modified data bases covering a period of nine months. In order to register the procedures and to identify the anomalies of the 4 kinds of incorrect procedures, a program as implemented. The obtained results were compared to those performed with another approach based on comparable corpora. The relationship anomaly did occur 146 times from which 76 different repetitions and of those, 27 were found in the corpus-based error approach. We did not find any occurrence of delimitation anomaly during the considered period. Also, related to the permanence anomaly we had found 5 occurrences which were coincident to the corpus-based error approach. Then, type anomaly had presented 18 occurrences, and all of them were recognized by the corpus-based approach. The thesaurus of MorphoSaurus Project has a consolidated coverage. So, after the analysis of the anomalies we concluded that the verification of procedures register contributes effectively to the thesaurus refinement.

ASSUNTO(S)

tesauros medicina - terminologia biomedical engineering engenharia biomédica - dissertações thesauri indexação medicine indexing engenharia medica sistema de recuperação da informação - medicina

ACESSO AO ARTIGO

Documentos Relacionados