Detection of Positive and Negative Sentiments in User Reviews

Abstract

Businesses are continuously looking for ways to improve their customer experience and please their target audience. The rise of digital technologies and social media has made it easier than ever for customers to express their opinions about products and services online. Customer reviews have always been crucial sources of information for businesses and organizations. However, with the rise of technologies, the volume of data generated by customer reviews has increased tremendously, and identifying and analyzing such a vast amount of data can be overwhelming, making it difficult for businesses to identify key insights and trends. To address this issue, we aim to develop a software system that can automatically analyze user reviews and determine the sentiment behind the text in Roman Urdu and English. The system will be designed to process large volumes of data quickly and accurately, enabling businesses and organizations to gain insights and improve their products and services. The project commences with compilation of a vast dataset encompassing user-generated reviews from a custom dataset which comprises of multiple predefined datasets. For sentiment analysis, our project has adopted the BERT (Bidirectional Encoder Representations from Transformers) model. The web development component involves the creation of an interactive and user-friendly platform using ReactJS, CSS and Flask. results. By using sentiment analysis on text, we aim to automate the process of analyzing customer reviews and provide businesses with accurate and timely insights into customer perceptions. Our system will not only help businesses to identify the sentiments expressed by customers but also enable them to identify key topics and themes that customers are talking about.