Machine Learning for Sentiment Analysis and Classification of Restaurant Reviews

Abstract

In this study, we performed sentiment analysis on restaurant reviews. To dine, we all go to restaurants with our loved ones and friends. However, we decide which restaurant to eat at before going there. Consequently, we select restaurants based on the reviews left by previous customers. A restaurant is a business, and the owner welcomes feedback from diners regarding their dining experiences so they may adjust their menu and amenities accordingly. Sentiment analysis is used to categories reviews as either good or negative, depending on whether they will have an impact on other people. Sentiment Analysis uses a machine learning algorithm and natural language processing to determine if a review is good or negative. It aids in the careful investigation of a scenario or service's good and negative effects, allowing for the suggestion of modifications. Both in terms of increasing consumer satisfaction and the caliber of the services offered, this might be advantageous. The user experience will be better as a result. This paper offers a study of key factors that distinguish between favorable and unfavorable assessments. This paper analyses various Machine Learning techniques that successfully analyze these reviews. In this study, we employed K-nearest neighbours Classifier, Logistic Regression, Support Vector Classifier (Gaussian NB and Multinomial NB), and Naive Bayes algorithm. A dataset from Kaggle was used in this study. Here, we will be able to observe which machine learning (ML) methods, can give more accurate results. We will also be able to evaluate how well an NLP can predict a sentiment from the given text. By illustrating how computers attempt to comprehend human sentiments, this article examines the possibilities. Out of everything, SVM produced an accuracy of 78 percent.