Analysis Of News Public Opinion And Evolution Under The Important Thought Of "Precision Poverty Alleviation"

The analysis of public opinion is a process,which is bringing relevant conclusions through a deep thinking on the public opinion of specific problems.In this thesis,the analysis of news public opinion means to explore the topic distribution of a large number of news reports and its evolutionary paths,and to explore the hotspot information hidden in large amounts of news by using of text mining technology.The common model for determining the distribution of topics is Latent Dirichlet Allocation(LDA)model.The difference in the number of documents over different periods was not taken into account,and the selection of the optimal number of news topic was performed only once before establishing the model.This can cause news topics to deviate from the actual.Therefore,before the model is established,we performs the selection of the optimal news topic separately in different periods.In addition,the ambiguous news topics offset the evolutionary path,so we should filter them before determining the path.The data used in this thesis are news reports related "precision poverty relief ".And the content of this thesis is divided into three aspects.Firstly,to explore the efficient way to obtain test data.Two common methods of data acquisition are introduced in this thesis,and the best method is selected by practical operation.Secondly,to study the distribution of news topic.Based on the relevant news reports,we establishes the text vector space with text segmentation technology,such as text participle,extracting feature words and weight assignment.Then we can establish LDA model in this vector space to explore the hotspot information hidden in a large amount of reports.Thirdly,to study the evolution of news topic.In this thesis,we use JS(Jensen-Shannon)divergence to measure the correlation between topics,and then determine the evolution of news topic over time,so as to predict the evolution trend of future news hotspots.It is found that the configuration of news template is an important part to acquire data efficiently,the selection of the optimal topic number is the key to determine the topic distribution,and the theme filtering can make the evolution of news topic more realistic.From the results of the analysis of public opinion,it is found that in recent years,the focus of news attention revolves around the Ten projects of "precision poverty relief".Tree poverty alleviation and photovoltaic poverty alleviation are widely implemented in the western mountainous areas.And this result can also contribute to the work of "precision poverty relief" conveniently,so that the work of "precision poverty relief" will be promoted more efficiently,and help people in poor areas to truly lift themselves out of poverty.