Research On Domain Ontology Model And Semantic Reasoning Of Animation Material

Currently,the retrieval technique of the mainstream information retrieval system is based on keyword syntax matching or content catalogue,which makes the computer rarely understand semantic contents and exists the phenomenon of missing some information that users really want and returning some information that users don't want.To do intelligent retrieval operations,the concept "Ontolgoy" is put forward to describe resources,enhancing the retrieval effectiveness essentially. Therefore,ontology-based information retrieval becomes popular. However,the ontology semantic expression is limited to DL(Description Logic),unable to express self-defined rules.So SWRL(Semantic Web Rule Language) comes into existence as the situation requires.Ontology and SWRL-based reasoning in the retrieval system guarantee retrieval precision ratio and retrieval recall ratio greatly.As the core work of the sub-project "Animation Material Semantic Retrieval System" in the 863 Program,this thesis focuses on the main contents as follows:Study the methodology for building ontology.Presently,there exists no systematic and complete methodology for building domain ontology. Hence,based on deep study on current incompact methodology,a new methodology integrated with the features of animation and suitable for building animation material domain ontology is brought forward,which emphasizes the assistance between ontology construction and semantic annotation.The experiments show that the methodology is in accordance with human's thoughts,and it owns strong operability,expansibility and self-adaptability.Therefore,the methodology has an important influence on ontology built in other domains.Build two kinds of ontologies:animation material domain ontology and animation task ontology named "Xing City Ji" according to the self-manufatured animated cartoon Travel of Xing City Popular Science after specifying the methodology.These two ontologies are core of Animation Material Semantic Retrieval System,for which they are the basis of the realization of semantic annotation module,index,semantic extension,semantic reasoning and semantic retrieval.Study ontology semantic reasoning mechanism.Reasoner Racer and reasoning engine Jess are deeply studied.More importantly,RETE reasoning algorithm is improved and an SWRL-based reasoning scheme is put forward after analyzing SWRL parser.The improved RETE algorithm speeds up reasoning process and enhance reasoning efficiency. According to the plots of Travel of Xing City Popular Science, SWRL-based reasoning rules are defined,with which forward reasoning are conducted。Then the reasoning results are feed back to the original ontology and the expected animation materials are retrieved.Implement Ontology Server for Animation Material Semantic Retrieval System and conduct a series of experiments on retrieval efficiency,recall ratio and precision ratio on basis of all of the above studies.The experiment results indicate that animation material domain ontology and "Xing City Ji" task ontology,and SWRL-based reasoning do ensure recall and precision.Hence,the whole system becomes more effective.