Central China Hotel Online Reviews Analysis Via Test Mining

In recent years,with the popularity of China's mobile Internet and the wide application of mobile phones,e-commerce develops rapidly,people are more and more used to consume on the Internet.Most of the people who go out for accommodation will choose to order hotels on the Internet,and will share their feelings of consumption on the Internet after consumption.This has resulted in the recent years,with the popularity of China's mobile Internet With the rapid development of e-commerce and the wide application of mobile phones,people are more and more used to consume on the Internet.Most of the people who go out for accommodation will choose to order hotels on the Internet,and share the feeling of this consumption on the Internet after consumption,thus producing a large number of text reviews,which contain rich information Some valuable information provides important reference for businesses and consumers.Consumers can judge whether to buy or not through these comments,and businesses can optimize through these comments to improve consumer satisfaction.However,due to the huge amount of data,there are many limitations in the method of artificial statistical analysis,so it is necessary to use the method of machine learning to deeply mine the valuable information in hotel text reviews in the era of information explosion.In this paper,through in-depth study of text data mining and online reviews of the relevant basic theory,using LDA theme model and text sentiment classification model,the author makes an empirical analysis of the online reviews of hotels in Central China on CTRIP Website,in order to provide consumers with purchasing opinions and provide reference for businesses to improve themselves.First of all,preprocess the text comments,transform the text data into the structural data that can be directly recognized by the computer,and get the key information that consumers value after word segmentation statistics word frequency:room,environment,clean,service.Generally speaking,consumers are satisfied with most hotels,but there are still some aspects that need to be improved;secondly,use LDA theme model extracts five comment themes:the overall environmental health of the hotel,service attitude,geographical location,price and the overall evaluation of consumers.Businesses and hotel managers can improve related facilities from these five themes to improve consumer satisfaction.Finally,based on the random forest classifier and naive Bayes classifier in the text emotion classification model,we classify the good comments and bad comments in the text comments.The accuracy of the random forest classifier and naive Bayes classifier are 95%and 87%respectively,which shows that the classification effect of the two classification methods is good.From the recall rate and accuracy rate,we can see that the effect of random forest classifier is better than naive Bayes classifier.At the same time,we prove that the machine learning method is effective for emotion classification and can be applied to text emotion classification.