Social media and networks are essential in today's world of communication technology. Because most people utilize social networking platforms to communicate their sentiments and emotions, each click generates vast data. As a result, we need particularly efficient ways to extract useful information from user data acquired from various social networking platforms such as Twitter. This research focuses on categorizing the different reviews offered by persons of various ethnic groups. The trials were conducted using real-time data obtained from a social media network. To classify opinions, sentiment polarities such as positive, negative, and neutral were used. Furthermore, the Convolutional Neural Network approach was utilized to conduct experiments, and the results were compared to other machine learning algorithms. The results reveal that the Convolutional Neural Network model can achieve accuracy levels of up to 94.47 %, 95.4 %, and 94 %, respectively, when evaluated using Phone, Laptop, and TV review datasets. The proposed framework can be utilized to develop a strong relationship between commercial businesses and their clients. In addition, this study discusses the different applications of sentiment analysis techniques, particularly in social networking and security.