Bacehlor's degree final project on Natural Language Processing.
Nowadays, with the increasing amount of user-generated data available online in micro- blogging services like Twitter, the human behavior of wondering what other people think has become an important field of study. Sentiment Analysis is a sub-field of Natural Language Processing (NLP) that tackles the issue of detecting the sentiment polarity of a piece of text, and is being used by a broad type of different businesses to extract insights of what people think about a product. The goal of this work is to perform a three-point scale {negative, neutral, positive} Sentiment Analysis on tweets from several editions of the BBC Proms, one of the biggest classical music festivals. We present a comparison of machine learning approaches and a state-of-the-art deep learning approach to classify Twitter data. In the experiments carried out, the deep learning approach, which is based on a CNN, outperforms the machine learning approaches achieving comparable results to the state-of-the-art.