Research On The Construction Technology Of Military Equipment Knowledge Graph

The rapid development of the computer industry and Internet technology has brought explosive growth of data to the human world,and these data often contain high utilization values.In order to be able to obtain valuable information from these data,researchers from home and abroad continue to propose new technology of big data management.Knowledge graph is one of the products derived from the era of big data.It is favored by academia and industry because of its powerful data management capabilities and visualization capabilities.At present,knowledge graph have been widely used in the fields of finance,corporate investment,and medical care,but they are relatively lagging in the military field.With the continuous improvement of the informatization degree of war,military operations have long been transformed from early single confrontation to system confrontation.System operations depend on the integration of many factors such as tactics,operational rules,weaponry and equipment,which puts forward new requirements for the informatization development of weaponry and equipment.This thesis studies the application of knowledge graph construction technology in the field of weapons and equipment,combined with the actual engineering needs of the joint training unit,realizes the intelligent management and efficient utilization of weapons and equipment data.The specific work can be divided into the following parts:Firstly,this thesis uses crawler tools to obtain high-reliability open source weapon data such as China Military Network and Jane’s Journals to establish the original weapon data set.At the same time,a dictionary of weapons and equipment was constructed based on the obtained structured data,and the unstructured text data was marked by the combination of the dictionary and the Chinese natural language toolkit Han LP for knowledge extraction of weapons.Secondly,this thesis has completed the extraction of military equipment knowledge.The knowledge extraction of weapons and equipment can be divided into two stages.In the stage of weapon and equipment entity recognition,based on the characteristics of weapon equipment data,this thesis constructs a Bi LSTM-CRF model that integrates multiple features such as word boundaries,part of speech,and radicals for equipment entity recognition.In the stage of entity relationship extraction,in order to solve the difficult problem of large-scale data labeling,and considering the relatively simple relationship of military data,this thesis labels the data based on the idea of remote supervision.In order to reduce the noise problem caused by remote supervision,the piecewise convolutional neural networks model based on the sentence attention mechanism is used for relation extraction to obtain entity triples.Finally,this thesis constructs the equipment knowledge graph and realizes the prototype system of military knowledge graph platform.The platform is developed based on Neo4 j,a native graph database,which enables users to quickly operate data without relying on the relatively complex CQL language.At the same time,the platform realizes functions such as entity query and intelligence analysis,and realizes the effective application of weapon and equipment knowledge graph.In addition,the platform can show the connections between data with a higher level of visualization.