Construction And Application Of Electromagnetic Knowledge Graph Based On Deep Learning

In recent years,with the continuous development of science and technology,related technologies in the electromagnetic field have also been greatly improved.Whether it is civil navigation or military electromagnetic countermeasure technology,it affects our lives every moment.However,at present,the acquisition of knowledge in the electromagnetic field is often obtained through inquiries on relevant web pages on the Internet or manual reading of relevant books.The knowledge obtained through these methods is too scattered,and it is difficult to form a relationship between knowledge and knowledge.It is difficult for researchers to quickly grasp electromagnetic knowledge and clarify the structure of electromagnetic knowledge system.Based on the information in the electromagnetic field,this paper constructs a knowledge graph system in the electromagnetic field,so as to solve the difficulty in acquiring knowledge in the electromagnetic field and break the problem that knowledge and knowledge in this field are difficult to form.Finally,a point-to-surface electromagnetic knowledge system is formed,which provides users with richer electromagnetic knowledge content.This article first obtains electromagnetic field data through data acquisition technology,and extracts knowledge from the collected data.Then build an electromagnetic knowledge graph system based on the extracted knowledge and apply it to practice.The main work is as follows:(1)Data collection and sorting in the electromagnetic field.Due to the lack of data sets related to the electromagnetic field for other scholars to study and analyze.This article uses acquisition technology to collect electromagnetic knowledge on the Internet on a large scale,and then through related processing,an electromagnetic field data set containing encyclopedia electromagnetic knowledge base,papers and patents is formed.(2)Based on an improved BILSTM-CRF entity extraction algorithm.For the task of electromagnetic entity extraction,this article first considers the impact of the semantics of the input text on the recognition of electromagnetic entities,so the input uses the BERT pre-training model for semantic learning;at the same time,considering the impact of model depth on feature learning,a deep BILSTM is used.Model;Integrating the above ideas,a BERT-BILSTM(deep)-CRF electromagnetic entity recognition model was formed,and it was verified on a real data set.(3)Relation extraction algorithm based on semantic embedding.Considering the particularity of knowledge in the electromagnetic field,for example,it is difficult to form relationships between entities in electromagnetic texts,and even in most corpus,there is even only one electromagnetic entity in a sentence,which causes the problem of difficulty in extracting relationships.Therefore,this paper adopts a method based on semantic embedding.The principle is that after semantic embedding,similar entities are often mapped to nearby positions in the vector space,and then whether there is a relationship is determined by judging whether the positions are close.(4)Based on the application of the electromagnetic field knowledge graph system.The graph system mainly includes the following applications such as: knowledge query,knowledge recommendation,knowledge navigation(block display),and the extended application of the system to assist reading in the electromagnetic field.Based on the above content,an electromagnetic knowledge graph system is constructed.According to the idea of software engineering,a requirement analysis,overall design,and system implementation and test analysis are carried out in real data.