Comparative experiments for multilingual sentiment analysis using machine translation

Sentiment analysis is the Natural Language Processing task dealingwith sentiment detection and classification. In the past few years, there has beena steady increase in the interest towards this task, for which different methods andresources have been proposed. Sentiment analysis has been studied in the contextof traditional media, but also the new social media. Nevertheless, the researchcommunity has concentrated less on developing methods for languages other thanEnglish.Motivated by this fact, the present article deals with the problem of sentimentdetection in three different languages - French, German and Spanish - usingthree distinct Machine Translation (MT) systems - Bing, Google and Moses, usingsupervised methods with different combinations of features. Our extensiveevaluation scenarios show that SMT systems are approaching a good level of maturityand can start to be employed to obtain training data for languages other thanEnglish and that sentiment analysis systems can obtain comparable performancesto the one obtained for English.