Representing Knowledge in Prototypes and Use Case in Biomedicine

Abstract

Diversity, complexity, and distributed nature of data formats and sources, and the extraordinary amounts of data, have been a great challenge for scientists. Data integration in practice offers at best locally valid integration methods, but with lack of desired flexibility and expandability much demanded for Web data sources. One dominant approach has been ontology-based knowledge representation. Ontology is a science of what exists, and theoretically for a given knowledge domain there can be an unlimited number of alternative ontologies in representing the domain. It is a worldview in which stable and ideally immutable classification of concepts and similarities among objects instantiated from them is constructed. Ontology as knowledge representation has gained significant momentum in science communities

however, due to its structural rigidity and lack of principled development and maintenance mechanisms, it leaves much to be enhanced to be used as a universal tool in accommodating varying knowledge entities, such as blog posts, not just scientifically proven evidence. Prototypes are a well-established research area in cognitive science and psychology for explaining human cognition. A prototype-based knowledge representation will more closely represent the ways in which humans construct knowledge in a computer-processable manner. After close examination of current such as ontology, conceptual space, and Frames, this thesis presents a prototype-based knowledge representation model implemented as an ontology as well as prototype knowledge base development approaches with biomedical examples, as well as a simple introduction to prototype learning. The prototype-based knowledge representation model is well positioned to address the drawbacks of the extant knowledge representation models, by offering ready expandability. The simple structure of the new model lowers the barrier for use by knowledge engineers, and fosters a greater freedom of expressions by them. Prototype-based knowledge representation is a new research venue that will open doors to many related research works.