Research On American Think Tank Text From The Perspective Of Keywords

The most important role of think tanks is to "support the government,enlighten the people,nurture talents,and exploit talents".In the information age,the relationship between human beings is getting closer and faster,which leads to the rapid changes of the international situation.The rapid understanding and accurate grasp of the international situation must be based on fast,accurate and reliable data analysis.It has become an important topic for scholars to study the text-based achievements of think tanks,explore the ideological trends of think tank experts in relevant fields,and explore the policy trends behind the text.Extracting effective information from a large number of unstructured texts and using natural language processing technology to research think tank texts is a hot spot at present,which is also the focus of this paper.This paper takes the think tank text-based achievements as the research object,takes the keywords as the breakthrough point,and realizes the keyword extraction of the think tank text by weighting the textrank algorithm,and applies the keyword extraction analysis to the analysis of the think tank text by means of visual display,so as to provide a new perspective for the construction and development of China's think tanks.The specific work is as follows:Firstly,the high-frequency words in the text are combined with the statistical comparison of Baidu filter thesaurus to extract meaningless words to build a stop thesaurus.Then,combined with the data and characteristics of think tank text,the weight formula of candidate keywords is redefined in the form of keyword + word length,keyword + word position,keyword + part of speech and other factors,and LV textrank algorithm is proposed.Secondly,by comparing with the traditional textrank algorithm and TF-IDF algorithm in terms of accuracy P,recall R,F value and combined with the comprehensive analysis of current affairs and political events,it is concluded that the algorithm proposed in this paper has obvious improvement in both extraction effect and extraction quality.Finally,taking the field of artificial intelligence as an example,combined with CO word analysis,gephi visualization and other means,the text research of think tank is promoted from the bibliometric unit to the content unit of keywords.Through visualization and comprehensive analysis methods,three research directions of American think tanks in the field of artificial intelligence are obtained,and the practical value of the research is verified.The comprehensive results show that this study improves the breadth and depth of think tank research,and improves the readability and accuracy of a large number of relevant think tank texts.