Managing Private Ontologies With A Bayesian Approach, For Time Critical Scenarios

Knowledge integration is fraught with problems. To overcome the "knowledge acquisition bottleneck", in time critical situations, knowledge-bases are being created utilising techniques that minimise interaction with a knowledge expert. This can lead to problems where errors occur within the created ontology, either with concepts, semantics or both, and can make integration a difficult task. A further difficulty for integration is that some domains have uncertain knowledge, which most integration techniques do not take into account. The integration framework discussed in this paper provides mechanisms for capturing uncertain concepts as well as the ability to capture the ontology creator's beliefs about concepts and semantics. These probability based concepts are used to assist the user when integrating the ontology with other ontologies. The Integration framework further permits the user to assign trust values for the ontologies, this allows the user to differentiate between an ontology created by a "knowledge expert" and those created by other, possibly less skilled, people. Current integration techniques provide assistance to users in the way of suggestions. This indicates that the user is the decision maker involved in selecting the correct merges and semantic relationships between concepts. This can cause problems as it relies on the user being an expert, who in time critical situations may not be available. This framework attempts to minimise the need for an expert in the integration process by providing probable solution-spaces that provide the user with alternate possible solutions that can be filtered to provide the best solution for the current situation.