Research On Construction Method Of Railway Route Selection Knowledge Graph

Chinese railroads go through a rapid construction and rapid development phase,along with the march of the railroad business,what is generated at the same time is a huge amount of knowledge resources,a large number of railroad case scientific data published in professional journals,this knowledge is expressed in natural language,identifying and extracting meaningful various line related named entities from the text,and constructing an association network according to the type and inherent relationship of the entities is an important information extraction method,which contains a wealth of knowledge in the field of line selection.In recent years,Natural Language Processing technology,artificial intelligence technology such as neural network,machine learning,and deep learning,and data visualization technology have flourished.Knowledge graphs can organically organize fragmented data by establishing correlation links between data,making the data more easily understood and processed by humans and machines,and providing convenience for search,mining,and analysis.This thesis innovatively introduces knowledge graph technology into the field of line selection design.Compared with the disorganized knowledge,the establishment of knowledge graph will make the knowledge more orderly and facilitate the collection,management,sharing and maintenance of knowledge.At present,the research on knowledge graph in the field of railroad line selection design is relatively scarce,and the graph involved in railroad field research does not cover more line selection knowledge,but only includes the part required for its research.Therefore,the knowledge graph in the field of railway route selection design introduced in this thesis covers more route selection knowledge,involves a broader range,has finer content,is more scientifically rigorous,and reflects more knowledge associations between railroad line entities.Using the graph database Neo4 j to improve the knowledge graph storage method and the application of knowledge graph,a knowledge query system is constructed based on the knowledge graph of railway line selection domain,which realizes the semantic retrieval of railroad line selection knowledge and will effectively support the acquisition of railroad line engineering knowledge and the discovery of universal relationship of railroad line entities.The specific research results are as follows:(1)Research on the acquisition of domain datasets and corpus annotation methodsBased on the crawler technology,800 literature abstracts of related topics on the CNKI were crawled to construct the data source of the knowledge graph in the railroad route selection domain.The corpus was annotated with entity types by using the wizard annotation assistant software,and the corpus with annotated entity types was annotated with unified BIOSE by writing code to construct a training corpus for learning named entity recognition models in the railway route selection domain.(2)Research on domain named entity recognition and domain entity relationship extraction methodsA training model for entity recognition in railroad route selection domain based on Bi LSTM-CRF model is established,and the Python language and Tensor Flow deep learning framework are used to achieve automatic recognition of named entities in railroad route selection domain.The model is also measured using three evaluation metrics and confusion matrix commonly used in machine learning,and the best F-value can reach 91.8%.Dependent syntactic analysis of the corpus using LTP tools achieves a new approach to discover and complement new relations of railroad knowledge entities.(3)Knowledge graph storage method researchA total of 21,891 railroad route selection knowledge triples were constructed by the knowledge extraction algorithm,and a knowledge graph storage method based on the graph database was realized.The collated knowledge triads are imported into the graph database Neo4 j by py2 neo,and the visualization of the railway route selection knowledge graph is realized.(4)Using front-end framework Vue and back-end framework Django and Java Script,developed a knowledge query system based on railroad route selection knowledge graph,designed knowledge management,knowledge query and other functions,realized retrieving and presenting data from the graph database,and presented the query results visually in the form of force-oriented graphs.