Sentiment analysis on yelp reviews

• Main Author: Wong, Hong Yong
• Other Authors: Sun Aixin
• Format: Final Year Project
• Language: English
• Published: Nanyang Technological University 2021
• Subjects: Engineering::Computer science and engineering::Computing methodologies::Document and text processing
• Online Access: https://hdl.handle.net/10356/148120
• Institution: Nanyang Technological University

Sentiment analysis, also known as opinion mining, is used to systematically identify, quantify, and study affective states expressed by someone towards a topic or phenomenon. In this project, sentiment analysis will be applied to determine the text polarity of reviews on various businesses left by Yelp users. Various supervised machine learning classifiers namely Naïve Bayes, Random Forest, Neural Networks, Support Vector Machine and Logistic Regression have been used to label reviews by performing text classification. These traditional classifiers will be pitted up against deep learning models and compared. They will be tasked with classifying Yelp reviews into five different ratings, and also three different labels according to the rating given by the user: - positive, neutral and negative. Negations have the ability to change the polarity of a given text and must be taken into account when performing sentiment analysis. The goal of this project is to analyze and observe the correlation between the use of negations in a Yelp review and the rating the user provided. The attention score of each token in the review text will be derived from two different deep learning models and compared.