Under the TCS iON RIO - 45 program, I had created a Sentiment Analysis Project.
I had cleaned the input paragraph text using NLTK module, so as to remove the unnecessary phrases and vectorized it to make it computer-understandable. Research was done on various suitable algorithms suitable for sentiment detection. Support Vector Machine (SVM) and Naïve Bayes were found to be the best. The cleaned text was passed into an instance of Multinomial Naïve Bayes, to get the sentiment score. Faced a few issues executing the plan, initially, but I did overcome them, eventually.
By applying the above algorithm, the generated model yields 86% accurate output and I am still working on it to improve its accuracy by training the model further.