Algoritmo basado en reglas de asociación para la extracción de relaciones no taxonómicas en corpus de dominio

Abstract

The identification of non-taxonomic relationships is a task that is carried out with learning and the creation of ontologies. Also, the manual construction of ontologies for experts and knowledge engineers is a costly and slow task, which is why it is necessary to create automatic or semi-automatic algorithms that speed up the procedure. In this research we propose an algorithm for the extraction of non-taxonomic relationships in an ontology of Artificial Intelligence (AI), evaluated through a data mining technique: association rules, which has statistical measures that determine the probability of occurrence between the concepts and the related connector verb. The experimental results indicate that 72 % of the relationships obtained in the algorithm exist in the ontology of AI.