Poblado automático de ontologías de perfiles académicos a partir de textos en español

Abstract

Resumen. Este artículo propone un enfoque para el poblado automático de ontologías de perfiles académicos a partir de los textos, expedientes curriculares y resúmenes, de publicaciones científicas en español. El enfoque utiliza reglas semánticas y marcadores lingüísticos para extraer los individuos de clase, relaciones y valores de propiedad. Una evaluación ha sido realizada con un conjunto de individuos gold standard, en términos de precisión y exhaustividad para el poblado de elementos ontológicos.Palabras clave: poblado automático de ontologías, reglas semánticas, marcadores lingüísticos, procesamiento de lenguaje natural (PLN).

A Proposal for Domain Ontological Learning

“…In these works are used techniques such as grammatical features extraction, named entities, morphological tagger and queries evaluation using WordNet with precision and recall. The academic profiles are explored in [6], where the authors carry out automatic population using curriculums and Spanish papers abstracts; they use a gold built for domain experts in evaluation step.…”

“…Their classes, relationships, constraints and axioms define a common vocabulary to share knowledge [5]. The ontology construction process can be manual or automatic, if it is automatic, the process is called "Ontology learning" and include: relevant terminology acquisition, synonyms extraction, concepts formation, hierarchical organization of elements, relationships learning, properties, attributes, with their rank and domain, hierarchical organization of relationships, instantiation of schema axioms, and arbitrary axioms definition [6]. In the manual construction process, a domain expert is necessary for model formalization; but generally, it is difficult to transmit their knowledge and the proper way to formalize it [7].…”