Sentiment Analysis Of Hotel Reviews Based On Dictionary And Machine Learning

With the rapid development of social networks and e-commerce,there are a large number of personal comments and reviews about products and service.These reviews are full of people's various emotions and great commercial value.Because vast data with wide sources is growing and changing all the time.If we deal with a mass of information with manual approaches,it not only costs a lot of time and manpower,but also it is ineffective.Therefore sentiment analysis is applied to analyze and process a large amount of reviews.Sentiment analysis is an important research area in the field of natural language processing.It has important research and practical application value.The paper mainly studies the sentiment analysis based on customer reviews in the hotel domain,which explores the attitudes of users on hotel services,etc and helps the hotel improve its service.The main work is as follows:(1)Based on the general Chinese sentiment dictionaries and using the hotel reviews as a corpus for the dictionary extension,the sentimental dictionary based on hotel reviews is built.The common dictionary and the dictionary built in the paper are respectively applied to classify the same reviews and the classification results are compared.The results show that the accuracy of the dictionary built in this paper is 76.5% and 80.4% in the positive and negative classification,which is better than the classification effect of the general dictionary.(2)Since the word vector representation method fails to consider the importance of the words in the texts,so this paper combines feature weights with word vectors and proposes weighted word vectors.The ordinary word vector and weighted word vector are used to represent text respectively and the classification experiment with SVM is carried out.The results show that the effect of using weighted word vectors to represent texts is better than that of ordinary word vectors,and the accuracy is respectively 85.2% and 81.6%.(3)The sentimental dictionary is combined with the machine learning method for sentiment analysis about hotel reviews.The training set is constructed by using the dictionary proposed in this paper and the weighted word vector is used to represent the text.The classifier is implemented and trained by two different algorithms,Naive Bayes and Support Vector Machine.The effect of the classification is verified by the test set.The results show that when classifying different numbers of reviews,the classification effect of SVM is better than that of Naive Bayes.When the number is 2000,4000,6000 and 10000,the accuracy rate of SVM is 4.7%,5.9%,7%,and 7.6% higher than that of Naive Bayes.