Design And Implementation Of Water Conservancy Public Opinion System Based On NLP

China is a flood-prone country,the occurrence of floods often brings a lot of adverse effects on people.In recent years,with the development of the Internet,it is more and more convenient for people to publish news and opinions related to water conservancy on the Internet.However,some news or opinions are often false and may even bring negative effects to the society.The existing public opinion system is universal,and there is no public opinion analysis for specific water conservancy industry.Meanwhile,the research on public opinion of water conservancy mainly focuses on theory.Therefore,it is feasible and effective to use emotion analysis and keyword extraction in NLP technology to monitor the spread of negative news to maintain social stability.This thesis aims at further research on emotion analysis and keyword extraction in water conservancy news public opinion analysis to provide a better monitoring scheme for water conservancy public opinion.This thesis mainly carries out the following research:In water conservancy public opinion sentiment analysis,a Bert-Bilst M news text classification model based on complete sentence segmentation is proposed to solve the problem that BERT directly truncates lost text information in long text,which can ensure the semantic integrity of text.In addition,compared with LSTM,BiLSTM takes into account contextual information and can extract deeper semantic information from text.The experiment proves that the F1 value of the model proposed in this thesis is 89.83%,which is better than the traditional Bert-LSTM in the accuracy of news public opinion identification.In terms of keyword extraction of water conservancy public opinion news,TextRank algorithm has universality in keyword extraction and does not involve word location information.This thesis proposes a multi-feature TextRank keyword extraction model of water conservancy news.In view of the universality of TextRank when extracting keywords,a dictionary of keywords in the water conservancy industry is maintained,and the existing keywords in the dictionary are endowed with greater weight and integrated into the algorithm.According to the word location information,the keywords appearing in different positions of the article are given different weights and fused into the algorithm.The experiment proves that the algorithm proposed in this thesis has a F1 value of 61.09%,which can obtain more accurate keywords when extracting water conservancy news keywords compared with the traditional TextRank algorithm.In the aspect of system design,Python is used for data acquisition,public opinion information monitoring and keyword extraction.The processed data is stored in mysql database,the server uses SpringBoot framework to provide public opinion data interface,and the front end uses React framework to process and display data.A complete water conservancy public opinion monitoring system is realized through multi-dimensional public opinion data display.