IRADABE: Adapting English Lexicons to the Italian Sentiment Polarity Classification task

Interest in the Sentiment Analysis task has been growing in recent years due to the importance of applications that may benefit from such kind of information. In this paper we addressed the polarity classification task of Italian tweets by using a supervised machine learning approach. We developed a set of features and used them in a machine learning system in order to decide if a tweet is subjective or objective. The polarity result itself was then used as an additional feature to determine whether a tweet contains ironical content or not. We faced the lack of resources in Italian by translating (mostly automatically) existing resources for the English language. Our model obtained good results in the SentiPolC 2014 task, being one of the best ranked systems.