The growth of microblogging services has expanded exponentially in recent years for mining user opinions. Sentiment analysis was applied to classify Twitter posts with video game titles as keywords. An analysis of the blog history, words and sentiments associated with the blog can help reveal whether the particular game is ‘violent’ and stress inducing or ‘non-violent’ and benign. An application was developed to collect and clean data. Naïve Bayes algorithm was applied to the cleaned data to determine the polarity of the words on the data to come to a conclusion whether, based on the words of the tweet, the particular game could be classified as ‘violent’ or ‘non-violent’. The results of the algorithm are analysed for accuracy, precision and recall. Deep learning models are discussed for use in future to improve accuracy.