Description Logic Ontologies Construction For The Semantic Web

As an essential research topic in computer science and Artificial Intelligent (esp. in knowledge representation and knowledge inference), Ontologies have been applied widely in many fields such as Semantic Web, semantic computation, search engine, knowledge engineering, software engineering, pattern identifica-tion, natural language understanding and so on.Semantic Web is extended from the existing Web, and it aims at making Web information apprehensible for computers to deal with, while ontologies de-scribe the semantics of Web information, and thus makes the inference of ontolo-gies possible. However, ontologies should be constructed before being used, as when there is any problem (such as antinomy, inconsistency and redundance), the inference of ontologies cannot be executed. Therefore, how to use ontologies language to contruct ontologies is the key to realize semantic Web.By analyzing the existing methods and the key problems of description logic ontologies construction, this dissertation focuses on the the non-standard infe-rence and the finite basis of description logics, the completeness of description logic ontologies, so as to study how to construct sound, unredundant and com-plete Web Ontology Language (abbr. OWL) ontologies, and provide theoretical basis and technological support for the realization of semantic information search.The main researches of this dissertation are as follows:(1)The non-standard inference of description logics.With the main research task on Least Common Subsumer (abbr. LCS) and Most Specific Concept (abbr. MSC) inference is studied. The non-standard infe-rence of description logics is mainly used to assist knowledge engineers to con-struct description logic knowledge base (also called ontologies). The research al-so covers the non-standard inference such as the MSC for description logic with n-ary existential quantifier, the MSC for description logic with transitive roles and existential restrictions?L+ and the LCS and MSC of hybrid terminological cycles in description logic?L w.r.t. descriptive semantics.Based on the fact that the non-standard inference of description logics are unable to deal with the insufficiency of n-ary existential quantifier, a new?L(n)description graph is put forward. And the inference algorithm of approx-imating MSC of description logic?L(n) is presented by using description tree and description graph, and the correctness of inference algorithm of approximating MSC is also proved by using?L(n) description trees embedding and?L(n)description tree and description graph homomorphism.Based on the fact that the MSC inference of description logics is unable to deal with the transitive roles and existential restrictions, a new?L+-description graph is put forward.Based on one of the description logic terminological cycles—description semantics (three types of semantics of terminological cycle:greatest fixed-points (abbr. gfp), least fixed-points (abbr. lfp) and description semantic), the grammar and description semantics of sL hybrid terminology cycles is presented, and the sufficient and necessary condition for the existence of the LCS and MSC infe-rence of hybrid terminological cycles in description logic?L w.r.t. descriptive semantics are presented by using TBox-completion and description graph. And the correctness of reasoning algorithms is proved. It is also proved that the LCS and MSC reasoning of hybrid terminological cycles in sL w.r.t. descriptive se-mantics can be computed in polynomial time.(2) The completeness of description logics ontologies.Based on the completeness of ontologies in constructing description logic ontology, the application of attribute exploration algorithm in constructing de-scription logic ontology is studied, and the insufficiency of the system that re-quires domain experts to possess all the domain knowledge in constructing on-tology by using attribute exploration algorithm is analyzed, and the constructing methods of ontology under the condition that domain experts do not possess all the domain knowledge is also studied. The completeness of description logic on-tologies in description contexts is redefined and the incomplete context is pre-sented. And an algorithm of constructing the description logics ontologies under the condition that domain experts are unable to define the attribute implications between those attribute sets is constructed. This algorithm and domain experts can acquire implication knowledge from each other thus an ontology knowledge base can be constructed, and the ontology constructed by this method is proved to be a complete one.(3) The finite basis problems in description logics are studied.Based on the redundance of ontologies in constructing description logic on-tologies, the finite basis problems of description logics are studied. The finite ba-sis problems in Description Logics (DLs), especially the fundamentality and the current research progresses of the finite basis problems in DLs are analyzed. The attribute implication and the Duguenne-Guigues basis in the Formal Concept Analysis (FCA) are studied. Based on the existence of the Duguenne-Guigues basis in FCA and the works of F Baader, a new context, namely description con-text and the implications of description logic formulas are defined. It is proved that there exists a unique greatest fixed-points (gfp) model in the terminological cycles of the description logic system FL?. Based on the gfp-model, the existence of the finite basis (a finite set of implications) in the terminological cycles in the description logic system FLs is proved. Moreover, the soundness and the com-pleteness of the finite basis are proved, too. Such a finite basis provides the knowledge engineer with interesting information on the application domain de-scribed by the description context. The knowledge engineer can use these impli-cations as starting point for building an ontology describing this domain.