SALF: um algorítimo para indexação automática utilizando vocabulário controlado

Abstract

A thesaurus based method of automatic indexing able to assign controlled descriptors formed preferentially by phrases, is developed. The method is designed to index titles and abstracts of written documents. It also includes device to suggest description to the indexer in order to enrich the indexing and the specific vocabulary of the area. With the aim to test the method, a sample from the FONTE/CIN data base in energy, is indexed using the method. A comparative evaluation in relation to the manual indexing is carried out; the parameters investigated are: recall, precision, exaustivity, specificity, coincidence of descriptors and concordance of the high frequency descriptors. The results confirm the feasibility of the proposed method.