The Research Of The Data Label Generation Method In Battle Field Sample Space

Under the environment of integrated operations,new requirements have been put forward for commanders to effectively understand the battlefield situation.Faced with a large number of combatants,advanced tactical technologies and various types of new military systems,the task of efficiently understanding the battlefield situation has become more and more arduous.Therefore,the "label" tool was introduced to analyze the battlefield data and assist the commander in making decisions.The battlefield label provides an efficient way for query,use,management and mining data of battlefield data samples.So,the label generation of battlefield data is particularly important.This paper proposes two modes of label generation.One is a smart method supported by computers,and the other is done by humanware(experts).Recently,pattern recognition and intelligent computing have attracted much attention in the academic and engineering fields.Under the guidance of situation awareness,the research results of artificial intelligence and other cognitive technologies are applied to achieve a breakthrough in situation awareness.In addition,the natural intelligence of commanders is integrated into the label generation technology.Firstly,this paper analyzes the work of researchers in label generation,and the research ideas on situation awareness.It also expounds the basic knowledge of rough set theory and points out the defects of classical rough sets,and proposes ideas for the following improvements.Secondly,the principle of constructing a battlefield data label framework is the basis of the behavior.According to the construction process of the label system,the core tasks of "Task Label","Capability Label","War Situation Label" and "Resource Potential Label" are constructed.A deep level and high-precision analysis was performed on each label during the battlefield data label framework.Thirdly,according to the characteristics of the battlefield data,a sequential three-way classifier combining the neighborhood decision rough set model and local attribute reduction was proposed.Considering the complicated and changeable battlefield situation,in some cases,the data labels that was generated by intelligent algorithms can't describe the battlefield data well.So,the humanware service is introduced to understand the battlefield data and generate battlefield data labels.In this process,the Jaccard index was used to test the similarity of the labels generated by the sequential three-way classifier.Then,a reasoning framework for label generation in the battlefield was designed.Finally,the paper simulates the experiment according to the above ideas.Compared with other algorithms,it is proved that the sequential three-way classifier proposed in this paper improves the accuracy of data label generation.