A Sentiment Analysis Classification of Product Reviews through Convolutional Neural Networks (CNN)

With the advancement of technology, the behavior of consumers turned into E-commerce, especially when the pandemic started. Defining the success and credibility of e-commerce is important as it is becoming a trend, and one of the factors that can affect it is product reviews. Product reviews build trust and loyalty, which influence the purchase decision. These product reviews can be in textual form, star rating, and emoji. Star rating is the commonly used product review but causes rating inflation. That is why word-based rating eliminates this problem which gives more accurate reviews. In this modern day, we do not only convey our sentiments in the language we are used to but also in other formats, such as using both text and emoji. Since Convolutional Neural Networks have been drawing attention to their reduced effort in feature definition, it has been gaining popularity in text classification.

This leads the researchers to create a tool that aims to determine the performance measure in detecting the polarity of a text-based and text-based sentiment analysis with emoji using Convolutional Neural Networks. The researchers determined the system's performance using Confusion Matrix and derived its Precision, Recall, and F1-Score.