Gate-Based Rules for Extracting Attribute Values

Abstract

Automated ontology population is intended toenrich ontologies. The main entities of an ontologicalmodel are classes, subclasses, attributes—datatypeproperties—, relationships—object properties—, andinstances. Class instances are important to thescientific community, since some work is devoted toautomatically populating ontologies by using statisticalmethods, information extraction, and natural languageprocessing, among others. The problem is focused onidentifying and extracting attribute values of instances.Commonly, such values have a predefined type likenumeric, string, boolean, etc. The difficulty arises whenyou want to know which instance belongs to such values.In this paper we propose an approach based on NaturalLanguage Processing (NLP) and Information Extraction(IE) technologies for extracting attribute values. We usesyntactic patterns implemented on the GATE (GeneralArchitecture for Text Engineering) tool. The results areindependent of the application domain and they exhibitpromising values of recall, precision, and F-measure.