CUSTOMER REVIEW ANALYSIS - MULTI-LABEL CLASSIFICATION AND SENTIMENT ANALYSIS

In this paper we focus on the hotel sectors and help them process these huge chunks of data in the form of customer reviews and help them derive useful information. The data pre-processing involves the scrapping of reviews from different sites and storing them and also check the correctness of the regular expression of the reviews. Our modelling employed includes three machine learning algorithms namely Naive Bayes, Support vector machine (svm) and Logistic regression. These three models improve the accuracy of the model as well as its robustness. The main idea of using these models are that the reviews are labelled so that the hotel management need not waste loads of time reading all the reviews. Instead the important reviews can be arranged based on their polarity and the important topic discussed in the review can be highlighted. So that it is easy for the management to analyse both the positive as well as the negative reviews. Sentiment polarity is incorporated to arrange the reviews based on the sentiment the review establishes. This paper helps the world to properly analyse the feedbacks and the reviews given by the customers.The very thought of processing thousands of reviews makes us nauseous. Analysis of data is essential in this information world. People with the help of this data can make themselves equipped and also improve their performance from that of their previous experiences. In this project we have successfully established a review analysis system, which can be used to find the topic on which the review emphasis and make the admin gain useful insights from it. This recommendation system can also be used to other type of recommendation systems and also data that requires multi-label classification