Analysis Of Internet Public Opinion Of "Three Children" Policy Based On Text Recognition

According to the data of the seventh national census,China as a whole has approached the "moderately aging" society.Population aging is a basic national condition throughout China in the 21 st century.Implementing the three-child policy and supporting measures is an important measure to deal with the problems existing in China’s current population structure.Due to the rapid development of mobile Internet technology and the open,interactive and low-cost characteristics of weibo,weibo has become a forum for the public to discuss problems,and there is a large amount of comment data on weibo platform.Therefore,the efficient use of these comments data is particularly important,in-depth study based on microblogging public opinion field has the comment on "three children" policy data,not only the government can help understand the thought of "three child" policy,but also can help the government to effectively formulate corresponding policies and measures to promote the implementation of the "three child" policy.This thesis takes the comment data under the blog posts related to the "three-child" policy published by several official micro blogs with more than ten million followers as the data source,and uses Python to write a web crawler to capture the comment text data of the above blog posts from May 31,2021 to October 31,2021 from Sina Weibo platform.Data cleaning,Chinese word segmentation and word removal are preprocessed for this part of text data.In this thesis,the preprocessed text data are analyzed as follows: First of all,feature analysis is carried out.By calculating the word frequency of the data,word cloud map and TF-IDF model are drawn to conduct preliminary analysis on the data;Secondly,the LSTM model and Text CNN model are trained with the open source text data set with emotion annotation,and some of the above preprocessed text data are manually annotated.Based on the trained LSTM model and Text CNN model,this part of the annotated text data is tested and the two models are compared.Finally,LSTM model is determined to be the better model and sentiment classification of all preprocessed text data is carried out through LSTM model.Finally,LDA theme model is adopted to conduct theme analysis on the comment data and study the topics that the public focuses on during the discussion of the "three-child" policy.Based on the above research,this thesis draws the following conclusions: First,among sina Weibo users’ discussions on the "three-child" policy,negative emotional comments account for 70%,while positive ones only account for 30%.From this data,it can be seen that most people have a negative emotional tendency towards the "three-child" policy at present.Second,in the public discussion on the "three-child" policy,the main topics of discussion focus on the following aspects: supporting policies,childcare costs,education,women,employment,etc.At the same time,there are some people express support on the policy.Based on the above conclusions,this thesis puts forward the following suggestions:First,relevant departments should actively guide public opinion through various ways such as Internet,TV and offline face-to-face,explain the current problems of population aging in China and publicize the important role of "three-child" policy in dealing with population aging;Second,all related departments to work together,as soon as possible to develop,and implement a series of effective,high feasibility and broad coverage of supporting policies and measures to these policies and measures need to involve the public in the "three children" policy concerns on all aspects of the problem,including education,female employment,child care costs and so on various aspects,do the public "can have three children,dare to have three children".