Research On The Construction Method Of Ontology Knowledge Base In Maternal Health

With the opening of the second child policy and the improvement of quality of life,people pay more attention to the health status of pregnant women during pregnancy and puerperium.By scientifically organizing knowledge and providing health services in the field of maternal health can effectively ameliorate the illness during pregnancy and childbirth.It can also actively protect the health of pregnant women and babies.Because ontology can effectively express,query,reason and share knowledge,and resolve the semantic conflicts of knowledge in different fields,this thesis introduces ontology in the field of maternity,and uses ontology to describe the maternal health information.At present,most of the ontology construction is manual construction,which takes time and effort,and is difficult to update and maintain dynamically.Therefore,this thesis focuses on the semi-automatic construction of the ontology in maternal health,that is,the extraction of information in the field of maternity,mainly including the extraction of ontology concepts and the extraction of relationships between ontology concepts(taxonomic and non-taxonomic).Based on the extracted results,combined with expert opinions,the construction of the ontology knowledge base in the field of maternal health was completed.The main research works of this thesis are as follows:(1)Propose a TFIDFEP method combining improved TFIDF algorithm and information entropyThis field has the characteristics of complex concepts and inter-concept relations,standardized terms,high accuracy,and low depth of natural language analysis technology.Compared with the extraction method of linguistics,the statistical method has higher flexibility and applicability.This thesis analyzes common statistical methods based on domain consistency,domain similarity,and TFIDF;then the TFIDF method is improved,and the TFIDFEP method is proposed in conjunction with information entropy,and the domain set is obtained by calculating the weight of the concept in the domain;the experimental results show that the proposed method has higher accuracy,recall and F-value,and can extract more and more accurate concepts of the field of maternal health.(2)Propose a method combining k-means algorithm and ant colony clustering algorithmIn this thesis,by studying the method of extracting the common relationship between concepts based on clustering,I propose a method of combining K-means algorithm and ant colony algorithm,select VSM to represent text,and calculate the similarity between concepts,and use ant colony algorithm to cluster concept terms K-means algorithm hierarchically clusters the terms,and uses the term with the highest similarity as the class label.This method can better achieve the automatic acquisition of the classification relationship between the concepts in maternal health.(3)Propose a method of combining association rules with VF * ICFFor the extraction of non-taxonomic relations,the non-categorical relations are often annotated by analyzing the syntactic structure and dependencies.The identification of verbs is the most critical task in this process.In this thesis,association rules are used to extract concept pairs,and VF * ICF is used to extract domain verbs,and finally the construction of triples is completed to better represent the non-taxonomic relations between concepts in the field of maternal health.(4)Build ontology knowledge base in maternal health.Preprocess the information in the field of maternal health,such as segmentation,tokenization,POS tagging,etc.Then,extract the concept terms and the relationship between concepts to lay the foundation for ontology construction;combine the suggestions of domain experts,use protégé tools to create classes and attributes,relations and examples,use SWRL language to compile disease diagnosis rules,construct the knowledge base of maternity,and store the ontology in OWL format in the MySQL database.