Sentiment analysis is the method of analyzing the emotions depicted in text-format such as tweets. There are 3 sentiment classes—positive, negative, and neutral. Sentiment analysis is done in order to analyze the customer reviews for a business and detect the emotions like happy, sad, angry, and upset. 20 Research papers were analyzed and studied for finding the details of techniques, datasets, algorithms, accuracy, and research gap. This study provided us with the prospective scope of advanced machine learning models like CNN with LSTM for classifying sentiments across a variety of topics accurately, including environmental sustainability, financial discourse, social issues, and many more. Urdu tweets and emoji embedding’s were analyzed using Decision Tree and CNN that achieved the highest accuracy of 95%. ConvBiLSTM model has achieved 91.13% accuracy on the Twitter dataset. Finally, this study covers various research gaps included in social networks’ sentimental analysis and areas for improvement in future work.