Muhammad Nasir
M.Sc. Candidate
Co-Supervisors: Dr. Orland Hoeber and Dr. Joerg Evermann

Department of Computer Science
Tuesday, November 13, 2012, 1:00 p.m., Room EN 2022

 

---

Abstarct

 

Ontology alignment research primarily focuses on creating state-of-the-art mapping algorithms. As ontologies are being commonly used to represent complex domain knowledge, their size can grow easily. As a consequence, the mapping algorithms produce a large number of mappings that ultimately the data manager has to view, manipulate, and validate. Visualization of these mappings usually results in a significant degree of visual clutter. In this thesis, we improved the visualization and perceived categorization of ontology alignment mappings using a modified edge bundling algorithm. The algorithm aids the users in performing better in ontology mappings tasks such as mapping verification and addition. A controlled user study was conducted to evaluate the benefits and drawbacks of the application of the bundling algorithm. The study verifies the performance gain resulting from application of these improvements for the mapping verification tasks. However, for the addition task mixed results were found.