Unfair reviews detection on Amazon reviews using sentiment analysis with supervised learning techniques

Abstract

Reputation and trust are significantly important and play a pivotal role in enabling multiple parties to establish relationships that achieve mutual benefit especially in an E-Commerce (EC) environment. There are several factors negatively affecting the sight of customers and sellers in terms of reputation. For instance, lack of credibility in providing feedback reviews, by which users might create phantom feedback reviews to support their reputation. Thus, we will feel that these reviews and ratings are unfair. In this study, we have used Sentiment Analysis (SA) which is now the subject generating the most interest in the field of text analysis. One of the major challenges confronting SA today is how to detect unfair negative reviews, unfair neutral reviews and unfair positive reviews from opinion reviews. Sentiment classification techniques are used against a dataset of consumer reviews. Precisely, we provide comparison of four supervised machine learning algorithms: Naïve Bayes (NB), Decision Tree (DT-J48), Logistic Regression (LR) and Support Vector Machine (SVM) for sentiment classification using three datasets of reviews, including Clothing, Shoes and Jewelry reviews, Baby reviews as well as Pet Supplies reviews. In order to evaluate the performance of sentiment classification, this work has implemented accuracy, precision and recall as a performance measure. Our experiments' results show that the Logistic Regression (LR) algorithm is the best classifier with the highest accuracy as compared to the other three classifiers, not merely in text classification, but in unfair reviews detection as well.