Research On Ontology-based Uncertain Knowledge Management

The knowledge economy age is upon us. Now, knowledge is endowed with the new role as the refill of natural resources, capital and labor, and becomes one of the primary driving factors for social development. Obviously, the management mode and theory for knowledge are urgently required in accord with the knowledge economy. For organizations and enterprises, knowledge is coming into being as a strategic resource with great economic values. In knowledge engineering, ontologies provide an effective means to describe the concept model and structure domain knowledge. With this background in mind, the thesis is mainly dealt with ontology-based knowledge management, especially in open uncertain environment. Several issues are addressed as follows:(1) Representation and query of uncertain knowledge based on ontologyThe environment (e.g. web, markets), in which organizations and enterprises are situated, is often open, dynamic and distributed. The knowledge related to this context is often uncertain, so that it can't be easily represented by traditional ontology modeling methods and modern description logic-based ontology languages because of their inner deficiency. In order to overcome this problem, we extend modern description logics by introducing probability to describe the dependence relations between concepts in domain and Bayesian Network to infer the relations. We also extend the knowledge query mechanism and provide support for uncertain knowledge.(2) A semi-automatic method for ontology acquisitionIn traditional knowledge engineering, knowledge is mainly acquired manually, which results in not only low efficiency and large cost, but also some serious inconsistency and various errors. We propose a semi-automatic ontology acquisition method, which attempts to overcome the unsurpassable knowledge bottleneck that has been further aggravated with the advent of web. The method provides general supports to learn the shallow ontology from various sources such as web pages and text, and validate it by domain experts to obtain the share domain model. Several problems are emphasized: how to crawl the text corpus based on ontology, how to...