Study On The Construction And Application Of Knowledge Graph In Chinese Bird Field

In response to the problems of large data volume,complex heterogeneity,and poor correlation of texts on the Internet,knowledge graphs can provide a new way of thinking for digital research of information in the field of birds,and use knowledge graphs to provide convenient knowledge services.Generic knowledge graphs have problems such as shallow knowledge depth for the bird domain.They can be used to fuse and organize fragmented texts by constructing bird knowledge graphs to convert heterogeneous information into structured information at a small cost and realize the unified organization and representation of domain knowledge.Ontologies can effectively organize and represent concepts,attributes,and relationships in bird knowledge,so this thesis constructs BKG4DK(Bird Knowledge Graph for Domain Knowledge)based on bird domain ontology and designs a deep learning-based knowledge extraction model for unstructured text to realize the content expansion of BKG4 DK.Finally,we carry out the research of intelligent services based on Bird Knowledge Graph for Domain Knowledge in China to meet the knowledge integration needs of users.The main work of this thesis has the following aspects.(1)Chinese bird domain ontology construction.The ontology can represent the concepts,attributes,and relationships in the domain knowledge in a general way.After determining the target of ontology construction,a seven-step approach is adopted for ontology modeling,which mainly includes determining the scope of the ontology domain and sorting out the core set of concepts and attributes.The ontology construction is completed using the Protégé ontology construction too based on the ontology modeling.Finally,the ontology is mapped to the knowledge graph according to the relationship existing between the ontology and the knowledge graph to realize the basic content filling of BKG4 DK.At the same time,the bird ontology can provide help for the knowledge inference service in the later knowledge graph intelligence service.(2)For the unstructured bird text,we design a deep learning-based knowledge extraction method to obtain the triad information embedded in the text through the joint extraction of entity relations to expand the scale of BKG4 DK.To address the problems of traditional annotation methods and combine the corpus features,we propose the ME+R+SOBIE annotation method which is compatible with the text features,and simultaneously annotates the entities and relations in the text to improve the annotation efficiency and annotation quality.On this basis,the Bird-Data corpus is constructed,and BERT,Bi LSTM,and CRF models are fused to train the bird entity recognition and entity-relationship extraction models to automatically extract entities and relationships in the text.To verify the actual performance of the model,different models are used to conduct comparison experiments,and the experimental results show that the model outperforms other classical models.(3)After determining the demand for knowledge services in the field of birds,research on intelligent services based on the Chinese bird knowledge map is carried out,among which there are knowledge management services,knowledge query services,and knowledge reasoning services based on BKG4 DK to realize the vertical application of the knowledge map.Among them,knowledge management based on BKG4 DK realizes knowledge acquisition and storage applications through knowledge acquisition and knowledge storage;knowledge query realizes knowledge retrieval and relational query applications based on Neo4 j graph database,and designs knowledge question and answer model to realize intelligent question and answer of simple questions;knowledge inference obtains new knowledge from existing entities and relations in the knowledge map by customizing knowledge inference rules.Realize the bird knowledge inference application.