Research On Sentiment Analysis Of Online Tibet Travel And Dietary Assessment Based On Bi-LSTM

With the rapid development of modern technology,the Internet has become the most important way for people to communicate.More and more Internet users express their opinions and opinions on what is happening around them through the Internet.For example,users buy products from online platforms and express their views on the products,or the government issues some policies.People can also express their views through the Internet.By collecting and analyzing these data,we can obtain the popular sentiment of the people.These sentiment-oriented data can help businesses better distribute the proportion of their products in the market.The government can further improve the policy based on the legitimacy of these data.Content and meet the actual needs of the people.Therefore,research on sentiment analysis of texts becomes more and more important.At present,sentiment analysis is mainly used in public opinion analysis,business product reputation and so on.Tibet is a famous tourist destination in China.Every year,many tourists from the mainland of China come to Tibet to play and taste the local food.Therefore,it is very valuable to study the sentiment analysis of mainland tourists on their tourist attractions and local food evaluation.Deep learning is to learn the correlation between data and apply this information to research tasks.The information obtained by these learning can be well applied in the fields of speech,text and images.At present,deep learning methods have achieved very good results in natural language processing,so they are favored by more and more people.Compared with traditional sentiment analysis,which relies too much on artificially built sentiment dictionaries and manual selection of features,especially the advent of the era of big data,the scale of data to be processed is getting larger and larger.Traditional sentiment analysis requires a lot of time and text classification It is difficult to further improve the results,and deep learning has made great improvements in these aspects.Therefore,this thesis studies the sentiment analysis of convolutional neural network(CNN)and bidirectional long-short-term memory network(Bi-LSTM),and combines them to propose sentiment analysis based on CNN + Bi-LSTM model.The main research contents completed in this thesis are as follows: 1)Using Python technology to collect comment texts about Tibet's tourism and catering industry on the Internet,20,000 data sets was constructed,divided into 13,000 positive emotional corpora and 7,000 negative emotional corpora.2)The Jieba word segmentation system is used to segment the collected data,and combined with CBOW in Word2 vec word vector technology to slide on these words according to a certain distance and train to get the word vector of each word.3)Based on the above work,a sentiment analysis based on Bi-LSTM model is proposed.The experimental results show that the accuracy of sentiment classification is effectively improved.4)In order to further improve the effect,the key information of CNN extracted text word vectors is combined with the text sequence features extracted by Bi-LSTM,and the sentiment analysis method of CNN + Bi-LSTM model is proposed.5)The SVM,CNN,Bi-LSTM and CNN + Bi-LSTM models were verified respectively,and the accuracy rate of the CNN + Bi-LSTM model reached 94%.