Ontology visualization and instance matches confirmation

Abstract

Joining multiple ontologies with the purpose of connecting existing knowledge can be done with the help of ontology matching tools. Part of this process involves establishing links between instances of classes from the paired ontologies. This process, referred to as instance matching, is done by algorithms that automatically generate the links between them. Since the suggested links might not be correct, people who are familiar with the involved ontologies can improve the outcome of this process by assessing the generated matches. The aim of this thesis is to investigate how to design a system which facilitates ontology understanding and involves users in the instance matching evaluation process. The tool is meant to be used by domain experts, with limited knowledge in Semantic Web technologies. To achieve this, a graph visualization of the ontology was developed and then evaluated during a two-phase iterative process. Later on, an instances-matched confirmation module was created and coupled with the visualization system. The user evaluation showed that domain experts are able to perform ontology exploration tasks similar to existing ontology visualization platforms and underlines people’s behaviour when they are exposed to large amount of data. The evaluation results also support the idea that users manage to improve the quality of the instance-matching by successfully using the match confirmation module.