Semantizar: um sistema de extração de relações semânticas

Abstract

Semantic relations are fundamental for understanding the nature of connection between conceptsin a domain, making representations of knowledge closer to reality. In thissense, the purpose of this paper is to present the web system called Semantizar. This system was developed to support the extraction ofsemantic relations for knowledge representation. For this, it uses as input conceptual bases, elaborated previously from academic documents and, the respective academic documents. As an output, the Semantizar indicates the sentence where the pairs of concepts occur, and the user validates the semantic relation between these concepts. A case study was carried out to verify the efficiency of Semantizar in the support to the extraction of semantic relations, the results point to a semantic enrichment of the conceptual basis using the Semantizar, because in the resulting representation all the concepts of the sample are related to the others. As a contribution, it is believed that Semantizar can act as an embryo for the automatic extraction of semantic relations, providing a dataset of semantic relations for the machine learning, in this way, it contributes with research that involves natural language processing.