Identifying Relation-gaps in Ontologies using TOPSIS

Abstract

Though many ontologies have a huge number of classes, one cannot find a good number of object properties connecting the classes in most of the cases. Adding object properties makes an ontology richer and more applicable for tasks such as Question Answering. In this context, determining the pair of classes that are most likely to have object properties between them becomes very important. We address the above question in this paper. We propose a simple yet powerful framework based on the popular multi-criteria decision making algorithm TOPSIS (Technique for Order Preference by Similarity to Ideal Solution), for identifying relation-gaps. In order to identify such relation-gaps (i.e class-pairs) the proposed system utilizes a mix of criteria derived from the ontology itself and from external sources such as Wikipedia text corpora. Our experimental results show that by means of these carefully chosen criteria and their corresponding weights, the proposed system yields promising results with respect to the precision of the relation-gaps identified.