Methodology for domain ontology development with human-centered design

• Main Author: Iqbal, Rizwan
• Format: Thesis
• Language: English
• Published: 2014
• Online Access: http://psasir.upm.edu.my/id/eprint/60488/1/FSKTM%202014%2020IR.pdf; http://psasir.upm.edu.my/id/eprint/60488/

Ontologies play an important role to envision the future of semantic web. Over the years, the practitioners of the field of ontology engineering to support the notion of merging different methodologies and techniques together for developing ontologies. This opens an avenue for new and different design ideas, making the ontology development process to be more effective and intuitive. However, through literature review it was found that most existing methodologies are extreme in their preference to include domain experts in ontology development process. There is a need to bring a balance in between the two extremes, so that a broader audience should be given the opportunity to participate and contribute in the ontology development process. Moreover, in existing literature, the notion of human centered design is neglected and there is a need to come up with methods which does not put burden on the audience for mastering them and should be easy to adapt for supporting ontology development. Furthermore, gaps related to immediate resolution to ambiguities, knowledge validation, manual selection of terms and ontology evaluation needs to be addressed. In order to bridge the aforementioned gaps, a methodology is proposed for developing ontologies using concept maps, with a focus on human centered design. The proposed approach is devised based on shortcomings found in existing literature as well as the lessons learnt while conducting preliminary study. The proposed methodology enables a broader audience to participate and contribute in the ontology development process, while exploiting natural language as the basis of ontology conceptualization. At the same time, it supports the notion of human centered design and integrates the element of learning, discourse and human feedback during the evolution of the ontology by exploiting concept mapping technique in different phases of the proposed methodology. Moreover, a term extraction engine (application) is also developed as a part of the methodology for automatically extracting the potential terms for concept mapping. For evaluating the proposed methodology, a study employing the mixed methods research design was conducted in collaboration with Quran domain experts (from University of Malaya and International Islamic University Malaysia). The mixed methods approach provided an in-depth understanding of the results and findings for answering the research hypotheses related to the proposed methodology. The results manifest that all the hypotheses are proven to be true, and the notion of human centered design is found to be successfully integrated in to the ontology development process. In addition, an ontology for the domain of Quran is generated as output from the study. Furthermore, an empirical investigation was performed to carry out ontology validation to ensure that the resulting Quran ontology is a true representation of the actual domain, as defined by Quran experts. The results of this investigation manifested the calculations for Closeness index (C) = 1 and Similarity index ( S) = 1, meaning that values of both measures indicate that the implemented ontology is an actual representation of the domain.