Research And Application Of Ontology-based Knowledge Base Construction Method

With the rise and rapid development of knowledge engineering and artificial intelligence,domain knowledge base has gradually penetrated into many fields for information management and knowledge discovery.Since ontology describes the relationship between concepts,it can infer the hidden relationships between concepts based on certain rules.Therefore,the use of ontology as a tool for knowledge representation and the construction of ontology knowledge base has become a research hotspot in the field of knowledge engineering.However,existing ontology knowledge bases only feed back the information in the knowledge base to users according to the existing knowledge systems or expert systems in knowledge bases,which lack the ability of analyzing problems so that can not provide users with valuable information or knowledge for unfamiliar problems.To solve above problems,some theories of the TRIZ innovation system were introduced.The TRIZ methodology contains tools for problem analysis,which can help the knowledge base to analyze user issues more intelligently and give more accurate search results.To reasonably and effectively use the TRIZ theory,taking the rice breeding field as an example,the structure of the rice breeding method ontology was designed,and the material-field model was introduced into the problem model of the knowledge base to help the knowledge base management system to analyze problems more accurately.Besides,to reduce the human participation and time consumed as much as possible in the construction of the ontology knowledge base,a rice breed identification method based on CRF was proposed for instantiation of ontology.The main research work is as follows:(1)Analyze the research status of ontology knowledge base both at home and abroad,and point out that the key problem of the current knowledge base is the lack of ability to analyze problems.At the same time,some theoretical knowledge in the TRIZ theoretical system was explored.By analyzing the characteristics of the ontology knowledge base,we find a more reasonable and effective method to combine the TRIZ partial theoretical knowledge with the ontology knowledge base.Besides,the development history of named entity recognition is reviewed,and the research status of named entity recognition based on CRF is analyzed.(2)Based on the TRIZ theory,taking the field of rice breeding as an example,the TRIZ functional analysis technology was used to design the structure of the rice breeding method ontology,and the ontology database was constructed using a seven-step method to complete the knowledge storage of the ontology knowledge base and expression.At the same time,a problem model is defined.The problem model introduces a material-field model in TRIZ,which can be used to functionally model user problems,more accurately analyze user problems,and improve the effectiveness of knowledge base retrieval.(3)In order to realize the automatic addition of examples of the rice breeding method ontology,the structure and characteristics of related patent texts of the rice breeding method were analyzed in detail,and based on which,a suitable text processing technique was used to extract the object attributes from the corresponding text.First,we use the patent claims as training data and apply SVM to mark the breeding method used for the patent text;secondly,we use information extraction technology to extract the patent's functional information from the patent abstract;finally,to more accurately extract the patent specification based on the stream information,a rice breed identification method based on CRF was proposed,which takes into account the characteristics of patent texts and the naming rules of rice,and effectively improves the recognition rate of rice names in patent texts.Finally,an experiment was carried out successfully to construct an ontology library of rice breeding methods and complete the implementation of a knowledge management system for rice breeding methods.At the same time,the validity of the proposed method for constructing the ontology knowledge base was verified.Experiments show that the proposed ontology knowledge base has strong problem analysis capabilities and can provide effective solutions and reference examples for user problems.