A data-driven dynamic ontology

Abstract

Valuable knowledge in every community is changed frequently. It often remains closely inside a community, even though it has huge potential to promote problem-solving in the wider community. Our research aims to increase the capability of communities in capturing, sharing and maintaining knowledge from any domain. We utilize an ontology, a shareable form, to collect, consolidate and find commonality inside knowledge. Most ontologies available these days were created by domain experts to fulfill certain domain requirements. However, in cases when domain experts are not obtainable or standard agreement within the domain is not available, such as in natural or herbal therapy domain, we propose that an ontology can also be extracted from existing knowledge-bases residing within the community. In order to achieve our aim, we design a data-driven dynamic ontology model. Our model consists of base knowledge creation and knowledge propagation phases. In the base knowledge creation phase, we define a general concept of capturing community knowledge from data into an ontology representation, rather than just transforming a specific data format into an ontology as found in existing studies. In our knowledge propagation phase, the dynamic community knowledge sources become the trigger of propagation. This is different from some approaches in existing studies, where the triggering event is an individual change inside the ontology and external data may not be the base source of the knowledge in the evolving ontology. We define the propagation feature with a novel delta script. The script is minimum yet complete to simplify and save knowledge sharing transportation resources. The evaluation result shows that the data-driven dynamic ontology with its propagation method not only delivers complete and correct semantics but also shows good performance in terms of operation cost and processing time.