Visual Research On Text Hotpot

In the era of big data,the speed from occurrence to development and finally to vanish of hot issues become more and more quickly.How to extract information that can make a guidance to reality from the complicated data becomes an important issue when the government try to explore hot issues and intervene in its development trend.For news data,due to the diversification of media and characteristics of news,the previous text representation methods can't grasp the characteristics of news well.Based on this,this paper normalizes the process of text visualization under the background of news text.Selecting news events involved in news text as the research object.This paper combined the correlation analysis model with network structure visualization model together,and proposed a set of standard in news text visualization process.The main contributions of this paper are reflected in the following aspects:1.This research focuses on the visualization of news hot spots,and standardizes a set of processes from news text processing to news hotspot visualization.2.In the process,according to the characteristics of the news text,we optimize the keyword extraction algorithm and visual layout of the network respectively to carry on the adaptability.This paper proposed algorithm based on tf-idf with TextRank algorithm which is adaptable to the expression of the news text.3.This paper put the method of combining correlation analysis and network structure to express news information,which can reflect the interaction between keywords and inner factors,as well as the changes of the hidden meanings behind the keywords more accurately,and it provide a good explanation for the hot migration.This paper applies the real estate news data in the first quarter of 2017 and integrates the above algorithms to realize the process to text visualization,and then verified the new method proposed in this paper.