Font Size: a A A

Topic Evolution In Radar Domain Based On Dynamic Topic Model And Information Entropy

Posted on:2020-09-22Degree:MasterType:Thesis
Country:ChinaCandidate:L L ChenFull Text:PDF
GTID:2518305732999009Subject:Signal and Information Processing
Abstract/Summary:PDF Full Text Request
The field of radar has been developed for more than 100 years and has developed many sub-fields and related technologies.It is widely used in social economic development and scientific research.As the time goes,many concepts,systems,technologies and application of radar have changed due to the changes in targets,environment,requirements and other related fields.The number of relevant research literature has also increase because of the breakthroughs and interdisciplinary developments in radar field.Faced with massive literature data,how to grasp the macro information and development trends of the discipline from the perspective of development is a big challenge for researchers and domain experts in the radar field.The evolutionary analysis of the topic of scientific literature can help researchers understand and discover the evolution of knowledge structures in the field to find hot topics and research trends quickly and comprehensively.However,there is currently no research on the application of topic evolution analysis to the field of radar technology.In order to provide objective reference and support for related research in the radar field,this paper first applied latent dirichlet allocation(LDA)-based discrete-time dynamic topic evolution model to analyze the knowledge evolution and topics evolution in this field.This paper analyzed the relevant English literature in the radar field from 1942 to 2018 in the IEEE database to analyze the topic evolution in the past 70 years.The main research contents of this paper are as follows:(1)This paper gives a detailed overview of the development,application and research status of the research on the subject evolution of scientific literature.We also discusses the significance of applying the subject evolution research to the field of radar technology.Several common methods for the evolution analysis of scientific literature topics are summarized in this paper.The basic theory,parameter estimation and model evaluation indicators of LDA and dynamic topic model(DTM)are emphasized.(2)A sampling method characterized by the cited quantity of the literature is proposed,and the sampling verification experiment is carried out.The result of original data model and the sampling data model are compared from the word frequency statistics,the content of topics and the topic strength trend.The verification experiment proves the credibility of the sampling method in this paper.(3)This paper uses the above sampling method to sample the literatures of radar technology domain in IEEE from 1942 to 2018,and model the sampled literature data by using DTM.In order to optimize the distribution of topic words,this paper applys information entropy to exclude interference words.The experimental results demonstrate the effectiveness of the method.This paper further visualizes and analyzes the above modeling results.A total of 77 topic were modeled in the sampled 38259 literature,and 65 valid topics were inferred.This article discusses the evolution of these topics from two dimensions:content and strength.In the evolution analysis of topic content,the word distribution of each topic is visually displayed in the form of a word cloud diagram,and the keyword changes of the topic are discussed.In the analysis of the topic strength evolution,this word combined with the artificial summary work in the field of radar technology and divides the evolution of the topic into ascending,descending and stable trends according to the overall situation of the curve trend.The analysis of the topic strength evolution provides objective evidence for the relevant review literature.Then,based on the strength value,the hot topics in this field are explored in each period.For example,we find the topic of "automatic target recognition","scene reconstruction","positioning navigation system","MIMO technology" and "super material" are hot topics in recent research in this field,and the attention of these studies will continue to rise.This paper objectively summarized the development history of research in radar field,discussed the knowledge structure in the field,and revealed the evolution and evolution trend of radar technology research.Our work can provide objective evidence for the existing review research and promote the development of innovative application in radar technology.
Keywords/Search Tags:Radar, data mining, text modeling, topic evolution analysis, LDA, DTM, information entropy
PDF Full Text Request
Related items