Ontological analysis of the wikipedia category system

Abstract

Copyright © 2018 by SCITEPRESS – Science and Technology Publications, Lda. All rights reserved We analyse violations of the transitivity principle of the Wikipedia category system, i.e. the situations where articles from a subcategory doesn’t logically belong to its parent category. The causes of the violation have been analysed on the base of ontological modelling methodologies such as OntoClean. We propose a new approach to automatically eliminating the violations. This approach is based on analysis of the relation of ontological dependence between categories. As a theoretical foundation of such analysis we propose a new deflationistic interpretation of the essential account of ontological dependence. The proof of concept has been evaluated on the category C:Mathematics. We are going to apply the proposed approach to derive a new large-scale domains hierarchy from the Wikipedia category system, and use it to provide BabelNet and DBpedia with fine-grained domain annotations.