The Key Technology Of Semantic Web-based Enterprise Knowledge Integration

With the coming of the era of knowledge economy, knowledge is becoming the key element of enterprise's survivement and development. Knowledge integration is an efficient approach to intelligentialize the enterprise. Knowledge integration is more and more important to enterprise. But too many theories and approaches of knowlege engineering are introduced in traditional knowledge integration research, which make the knowledge integration of enterprise difficult. The rapid devolopment of semantic Web opens up new opportunities for the research of knowledge integration of enterprise. It can be found that applying the technology of semantic Web to knowledge integration system is a "win-win" solution. Against this background, some key problems about constructing semantic Web-based knowledge integration system in enterprise are discussed.Research works are taken as follows:1) Aimed to the problems discussed in this thesis, the theory devolepments of knowledge integration, semantic web, ontology and Case-based Reasoning (CBR) are reviewed. The development and facing difficulties of ontology construction and ontology learning are generalized. Typical approaches of ontology integration are analyzed, and superiorities and deficiencies of them are summarized. The researches of ontology-based CBR at home and abroad are reviewed and the main deficiencies are analyzed. Then the overall research planning and the technical lines of the several key problems are put forward.2) In order to get around ambiguities in Chinese word segmentation and reuse a lot of domain knowledge in enterprise legacy intelligent systems, a novel approach for learning OWL local ontology from legacy intelligent system in enterprise is proposed in this thesis. On the basis of formal modeling source data and goal, the element correspondence between relational database schemas, tuple set and OWL ontology is analyzed, the OWL ontology learning approach and revelant automatic mapping algorithm with low time-complexity are proposed. The approach consists of two steps that are acquiring relational database schema and mapping relational database schema and knowledge item to OWL local ontology. Compared with existing methods, the approach whose data source implies more domain knowledge is more appropriate for actual engineering application. OWL ontology from legacy intelligent systems can be acquired automatically via a simple translation algorithm with low time-complexity instead of using a middle model or a lot of abstract learning rules, and numerous knowledge items about rules and cases (tuple set) can be reused as instances of OWL ontology according to certain priority. Validation of the approach is done by an application instance learning OWL local ontology from a legacy intelligent system in the wide of Tooling Man-hour Rationing in the enterprise wide.3) Advanced a semi-automatic constructing OWL domain ontology approach that is based on FCA and computing concept equation measure and concept inclusion measure. In the approach, constructing formal context is based on similiarity measure about name, structure and attribution. Concept clustering is based on concept lattice construction. Suggestions to enrich or amend concept hierarchy of ontologies are made automatically, which is based on computing concept equation measure and concept inclusion measure. The final OWL domain ontology can be construted by designers based on their own domain knowledge. Compared with existing relevant methods, here faces a new problem whose data source is servral OWL local ontologies and whose task is ontology mixing and constructing. The approach is combining FCA with similiarity measure, which comes better performance about difficulty of implement, complexity of lattice and semantic intensity. Validation of the approach is done by the evaluation of an experiment result.4) At the basis of analyzing that the semantic web and CBR can benefit from each other in integrating environment, a novel kind of semantic web-based case representation and retrieval arithmetic is proposed in this thesis. A RDF-based Case Web Markup Language (RCWML) that we define is applied for experience knowledge representation in order to flexibly integrate special case knowledge with general domain ontology and interoperate different cases of different case bases. Relevant case retrieving arithmetic can be more precise through combining the measure of domain ontology-based concept similarity assessment. The approach can get around the difficulty of strictly logic reasoning in Web environment and achieve effective integration of different case bases. An application scenario from Tooling Man-hour Rationing in the enterprise wide is introduced to evaluate our methodology.