• Analyzed academic and technical literature on data mining for sentiment analysis on a variety of text corpus. • Collected news article dataset, from GDELT online news services (http://www.gdeltproject.org/). • Conducted a user survey, for categorizing the collected news items, as violent and non violent. • Implemented Training using Python’s scikit-learn module, employing Decision Trees, Random Forests, Logistic Regression and SVM algorithms. The data was divided into training and test sets and td-idf vectorizer was used, to balance the usage of low frequency words in model training. • Implemented parameter tuning on the implemented algorithms, using training, validation and test sets. • The project resulted in a final model, having the accuracy of approximately 76 percent, in categorizing news articles on the basis of their violent and non violent content.