Research On Key Technologies Of Ontology For Intelligent Decision-making Applications

As a kind of knowledge management model,ontology has been widely used in artificial intelligence and knowledge engineering.It plays an important role in knowledge sharing,knowledge reasoning and intelligent assistant decision making.Especially in the field of intelligent decision making based on knowledge reasoning,it is necessary to model and manage the knowledge described in the form of Chinese documents.Ontology can formally preserve the semantic relations between terms and terms in a particular domain or task,provide a standardized and unified description of domain knowledge or task problems,and provide model support for knowledge sharing,reuse and reasoning.Therefore,it is necessary to introduce domain ontology and task ontology to build the corresponding knowledge model in the field of aviation command intelligent decision.However,the construction of domain ontology and task ontology is dominated by human methods,which is obviously time-consuming and laborious,and semi-automatic or automatic construction method has become a hot research topic.Supported by the intelligent decision-making project of aviation command,this paper mainly focuses on the semi-automatic construction of Chinese domain ontology and task ontology.Obviously,the semi-automatic construction technology of Chinese ontology involves two key problems,that is the extraction of terms and the relation between terms.Therefore,the key technologies of ontology for intelligent decision making applications studied in this paper include:Chinese terminology extraction,Chinese terminology relation extraction,Chinese domain ontology construction and Chinese task ontology construction.The main contributions of this paper are as follows:Firstly,a method for extracting domain terms based on text features and complex statistical weights is proposed?TCS?.The method first preprocessed the natural language Chinese domain documents,and then getting the candidate term w_ij through the coarse filter and considering the text feature and composite candidate terms w_ij statistics,calculate the comprehensive weight of WT(w_ij),and extract the final term from the candidate terms which WT(w_ij)value is greater than the set threshold.The experimental results show that the proposed method can effectively combine user dictionary,text features and statistical rules,and has a good effect on Chinese terminology extraction in specific fields and a high accuracy.Secondly,a term relation extraction method based on hybrid cosine similarity kernel function is proposed?MCSK?.This method can extract hierarchical relationships between terms by constructing the hybrid cosine similarity kernel function of the cosine similarity of sentence pattern semantic sequence in Chinese natural language documents and the cosine similarity of candidate hierarchical relational words,combining template rules and semi-supervised machine learning methods.Experiments show that this method can improve the shortcomings of using template rules and machine learning methods alone,and obtain higher accuracy and performance by using fewer manual templates.Thirdly,a semi-automatic construction method of domain ontology is proposed.The concrete steps of this method include:determining the research field,preprocessing Chinese knowledge documents,mining core terms,extracting and clustering the relations between terms,owl ontology structure and Protégévisual correction.The effectiveness of the method and the high task completion rate are verified by building an example and comparing it with other methods.Finally,a semi-automatic construction method and query algorithm for task ontology is proposed.The decomposition of tasks,dynamic IDEF3 model,formal description of task ontology and IDEF5 modeling of task ontology are discussed in depth.A semi-automatic construction method of task ontology and query algorithm based on task ontology are presented.An example is given to verify the effectiveness and high performance of the method.In order to verify the effectiveness of the proposed method,a prototype system of an ontology-based command and decision support platform?CDSBO?is designed and implemented.The simulation results of the system verify that the proposed method of extracting Chinese terms and terminology relationships,and the semi-automatic construction method of domain ontology and task ontology have high performance.It provides a powerful support for reasoning tasks of ontology in the field of aviation intelligent decision,and realizes automatic/semi-automatic inference generation from command intention to decision scheme.