Semiautomatização de relações em tesauros: uma proposta para o refinamento de relacionamentos semânticos a partir do tesauro Agrovoc

Thesauri are tools that contribute to information retrieval in information\udservices like digital data bases and digital libraries. Objective: This study aims to\udpresent a quantitative analysis on the refined sematic structure of the AGROVOC\udThesaurus and propose a semi-automated model for semantic relationship refinement\udof thesauri. Methodology: To carry out this study, we used data from the AGROVOC\udThesaurus represented in a refined SKOS-XL model. This data has been qualitatively\udaggregated in the classification of concepts by entity types as well as the hierarchical\udclassification of relationships done by Soergel. From this base, we conducted a\udquantitative analysis of the relationship types. Results: Results from the quantitative\uddata analysis showed that the refinement of the AGROVOC Thesaurus is not yet\udcomplete. Most Related Term relationships seem to have been refined, but hierarchical\udrelationships (Broader / Narrower) have not. Conclusion: In sum, this study\uddemonstrates that quantitative analysis can shed light on the structure of a thesaurus\udand indicate areas where improvements are possible.\u