Sentiment Analysis on Headlines

Determining emotion in a piece of text is a subset of sentiment analysis, a field which has largely focused on binary classification of text as positive or negative. Classifying sentences in one of six major emotion categories (anger, disgust, fear, joy, sadness, and surprise), however, can offer more insight and contribute to political polling, marketing, and stock market predictions.

In this study, we focus on the first of these six emotions, anger, and develop a model to predict whether or not a reader will feel anger after reading a headline. Our model involves NLP tools - stemming, named entity recognition, and part of speech tagging - and a bag of words model run on a SVM with a linear kernel and a Naive Bayes Classifier with Laplace smoothing. We developed experimental models to test on smaller subsets of data including: PCA, to reduce dimensionality of data, a synonym algorithm, to expand each headline allowing our model to classify with respect to more words, and a co-occurrence model, to score each headline based on co-occurrences of words with the word `anger’ in the New York Times archives.

We also built an emotion headline generation model with Markov chains to see if we can automatically create angry headlines. With these models, we hope to gain greater insight into methods that one can use to detect emotion within text.

This project was developed with Bharad Raghavan. See project here.