Research Hotspot And Frontier Discovery And Analysis Of Management Science Subject Based On Text Mining

In recent years,the number of related researches on management science has been increasing continuously.Faced with the increasing academic resources,it is of great significance for scientific research workers to timely and accurately classify the research fields of management science in China and grasp the buzz spots and frontiers of research.In previous studies,the methods of bibliometrics or co-occurrence analysis of key words in foundation papers are often used,which have relatively subjective limitations.Due to the limited work on preprocessing and model optimization,the results obtained in the few articles using text mining method are not convincing enough.In order to try to use text mining method to identify buzz spots and frontiers,and improve the rationality and objectivity of the results.This article selects in the management science fund project of national natural science fund "keywords + abstract" as a corpus of text.In order to improve the accuracy and scientificity of the results,in the third chapter we improve the algorithm to generate user dictionary in order to enhance the accuracy of the segmentation results.In chapter 4,buzz words and emerging words and their quantification methods are proposed to provide new ideas for identifying hot spots and frontiers of research based on the perspective of text mining.In chapter 5,we tune the LDA theme model and improve the calculation method of keywords.To improve the accuracy of text mining results,the text was pre-processed through four steps: word segmentation mode selection,term merging,word removing and word screening.By referring to the algorithms used by predecessors in unknown word discovery.We improved it by using the term obtained from the initial word segmentation as a unit instead of word,combined with mutual information and left and right information entropy and word frequency,to generate user word dictionary for term combination,so as to optimize the word segmentation results.Based on the concepts of research hotspots and research frontiers defined in this study,this study first gives the concepts of buzz words and emerging words and quantifies them to identify research hotspots and research frontiers and analyze their development and changes.The research found that among the fund application period 1993 to 2015,the research on the five topics of enterprise,economy,knowledge,resources and evaluation has been widely concerned.And the hot spots of fund project of adjoining two years have become increasingly similar.According to the results of emerging words,the research frontier of domestic management science research can be divided into five time stages,and the research frontier of this stage is consistent with the national policy and economic development situation of this stage.This shows that the fund will pay more attention to the research content in line with the current national conditions and situation.This study USES the LDA thematic model to divide the research fields of management science in China.Compared with the division of research contents in the field of management science in previous studies,this study obtained a more reasonable and detailed division of research topics through word segmentation results optimization,determination of the optimal number of topics in LDA,and calculation improvement of the correlation degree of subject terms.From the concluding fund project from 1993 to 2015,we can find 17 different research topics representing 17 different research fields of management science,including 10 hot topics and 9 frontier topics.Among these research topics,"China's economic growth and energy conservation and emissions reduction and monetary policy research","enterprise strategic competition and knowledge innovation ability research","the governance of the listed company and the company regulation research" and "industrial cluster and industrial upgrading,and intellectual property rights related research","human resource management and employee performance management research" are those which closely related to national policies,and are highly concerned in fund projects.They are hot and frontier topics at this stage.This study improves some limitations of previous studies,and selects "key words + abstract" as the corpus text in the research data of national natural science foundation of China.This paper discusses the relationship between research hotspots and research frontiers,and defines the characteristics of research hotspots and research frontiers from the perspective of text data.On the research methods,the text preprocessing phase drawn lessons from the new discovery algorithm,puts forward the preliminary segmentation using word rather than Chinese character,about combination of mutual information and the information entropy to generate according to different users of text data dictionary,the results obtained with this method has high accuracy and can save a lot of artificial processing time;This paper proposes the concept and quantitative index of buzz words and emerging words from the perspective of words to reflect the research hotspots and research frontiers,and the method to analyze their development and changes,which provides a new idea for mining the potential hot and frontier topics and their changes in the text data.Through the optimization of LDA subject model,through the determination of the optimal number of topics and the improvement of the calculation method of the correlation between topics and lexical items,a more accurate and reasonable division result of the research field is obtained.