Research Of Information Extraction Algorithm On Sentiment Analysis Of Household Appliances Enterprises Network Reviews

With the wide application of network information technology in many aspects,the comments and news information on the network are growing rapidly,and people are happy to send micro-blog,comment and news on the Internet.We can leave a message or comment on the Internet after we buy something on the appliances website.Through the emotional analysis of these reviews,we can find the defects and advantages of products and after-sales service,help businesses to improve products and services and help customers to buy the appropriate goods.These comments will produce massive explosive information.So much information can not be extracted effectively and make emotional analysis by means of labor.In this paper,we design and use some information acquisition algorithms to study the sentiment analysis of customer reviews after the sale of household appliances enterprises websites.The following is the specific content of the research.(1)sentiment analysis algorithm of review text based on RNN(recurrent neural network)In the design of RNN algorithm,the algorithm uses Jieba for data preprocessing,embedded embedding layer,LSTM neural network,algorithm parameters debugging,training and other techniques.Both the RNN(recurrent neural network)algorithm and the LSTM neural network are combined to improve the accuracy of the RNN algorithm.(2)sentiment analysis algorithm of review text based based on XGBoostThis algorithm uses the TFIDF algorithm to extract the features of the word,and then takes the extracted features as the input data set of the XGBoost.It simplifies feature extraction of the ordinary tedious text word vector,and I optimize the XGBoost algorithm by orthogonal verification and grid search.(3)sentiment analysis algorithm of review text based on light GBM combinationThe algorithm uses TFIDF algorithm to extract word features firstly,and then uses Logistic Regression algorithm to extract word features,and takes these features as input data of light GBM algorithm.The combination of Logistic Regression algorithm and light GBM algorithm is used to improve the overall algorithm effect,and optimization of the Bayes and algorithm stacking is used to improve the light GBM algorithm.