Ambiente para geração e manutenção semiautomática de tesauros

Abstract

Thesauri are among the diverse means of representing information as used by Information Retrieval Systems, which is considered to be a consolidated indexing language employed by professionals that carry out activities of organizing information. The flexibility for establishing new relations between terms, the hierarchies and the crossed references give that instrument a diversity of usage, reaching processes that range from indexing to effective recovery of documents. The production and maintenance of a thesaurus are intellectual activities with specific procedures to be followed. Among them are knowledge about documents produced in the subject area, comprehension of the used terms and the construction of concepts to explain those terms. It isexpected from the professionals in this field a flexible attitude to assimilate the changes and innovations which may be found in the indexing area, in the language itself, and in the usage of terms. This study aimed at the construction of a methodology and an environment to the generation and maintenance of a thesaurus within a semi-automatic way, through the use of natural language based on the technology of Computer Science and the theoretical scope of Information Science. Moreover, through the ordering concepts of literary, usage and structural guarantee, they incorporate the proposal of the guarantee that comes from the structure of the text itself. The environment made it possible toverify the present thesaurus as well as its representative potential. Thehypothesis was that key words from scientific articles could be applied in this process, since they represent both the literary and usage guaranties, for they are a privileged instrument in disseminating the scientific knowledge. Statistic calculi were made involving frequency and score standardised in the observation of the frequency of key words in titles, summaries, texts and articles of the reference list. Scientific texts of the electronic periodicals 'DataGramaZero and Ciência da Informação were used. The absence of an updated thesaurus in the area led to the elaboration of a Thesaurus in Information Science (TIS), from the existing thesaurus in Portuguese (IBICT), in the English language (ASIS) and in Spanish (CINDOC and DOCUTES). The environment pointed out the need of updating terms classified by degrees of relevance, considering the progress of the area.