Application Of NLP Technology In Agricultural Public Opinion Analysis System

With the rapid development of Internet information technology,the major news media regard the Internet as the main way of information transmission,and the Internet has become the preferred way for people to obtain news.The Internet has brought us convenience,but at the same time,it has also brought some problems.Because of the rapid speed of the dissemination of interconnected information,if the negative news or even rumors are not dealt with in time and paid no attention to,all kinds of false information will be disseminated quickly on the Internet,resulting in greater social impact.In order to avoid this kind of situation,it is urgent to use the Internet to monitor the Internet public opinion in time.Through the monitoring of public opinion,we can quickly find the recent news hotspots,timely guide the negative events correctly,and avoid causing greater social impact.Based on the national agricultural product traceability system,this paper studies the design and implementation of the agricultural public opinion system in order to better monitor the public opinion related to agricultural products and help the staff of the Ministry of Agriculture to carry out the public opinion early warning of agricultural products.In the research,based on the demand of the public opinion system of agricultural products,we selected the websites related to agricultural products news such as China Agricultural News Network and China Agricultural Products Network as the data source of the public opinion system.Word embedding and LightGBM are used to build a text categorization model for agricultural news and identify the news related to agricultural products.TF-IDF(Term Frequency-Inverse Document Frequency)and Ridge Regression are used to build text classification model to extract negative news.In order to get a faster understanding of news topics,an improved TF-IDF algorithm is used to extract news keywords.In order to help early Warners read news faster,text summary technology is used to compress news and extract news summary.The contents and innovations of this paper are as follows:Firstly,a text representation method based on word vector dimensionality reduction is proposed.The vector features of text are extracted by word vector dimensionality reduction.Word vector is applied to the classification of long text,which achieves better results than traditional TF-IDF in the task of agricultural news recognition.Secondly,in the task of keyword extraction,a keyword phrase detection method is proposed,which uses co-occurrence matrix and word co-occurrence graph to extract keyword phrases in news.Thirdly,an automatic text summarization method based on TF-IDF is proposed.This method uses the TF-IDF of words and sliding window to calculate the TF-IDF value of sentences.At the same time,it combines the location characteristics of sentences,and achieves better results than other text summarization methods.