A method for integrating ontologies, contextual variation and rules, and its application to reading material classification

[摘要] Rule-Based Systems (RBSs) are Knowledge-Based Systems whose knowledge is structured by rules. Our idea is to combine Contextual Ontologies (CO) with RBSs to synergise the advantages of both systems. We aim to make the improvements in two fields: (1) ontologies, and (2) RBSs. We aim to improve the accuracy and possibly other types of quality in (1) by thoroughly integrating contextual information into an ontology, rather than having it separated, and to improve RBS performance in (2) by adding contextual ontological processing into RBSs. ContextOntoRBS has been proposed, where concepts can be varied according to ”contextual concept-variants”, which can be abbreviated to “concept-variants”. We aim to achieve the improvements in both (1) and (2), independently. Reading Material Classification (RMC) has been chosen as the particular test-bed, and the availability of its dataset will be used to evaluate the improvements in (1) and (2) by proposing a new approach for rule-based RMC. RMC is a classification task where the input text is classified into the particular readability graded reading material. From the evaluation experiments, we do not claim that our proposed method is better than machine-learning-based RMC. ContextOntoRBS’s performance is just on a par with them. Rather than beating machine-learning-based RMC, we aimed to use RMC for showing that the inclusion of the “integrated contextual concept-variants” in ontologies and RBS provides the improvements over not including them. It can be concluded that by integrating contextual information into an ontology and into an RBS, the improvements can be obtained for both fields.

[发布机构] University:University of Birmingham;Department:School of Computer Science

[关键词] Q Science;QA Mathematics;QA75 Electronic computers. Computer science