Identificando sentimentos de texto em português com o SentiWordNet traduzido

Abstract

Sentiment Analysis is the find of study that examines opinion written in text. Researches in the field of sentiment analysis is often applied to english texts, and other languages like Brazilian portuguese are little explored. In this study it’s shown an implementation to semantic classifiers that analyze sentiments on texts written in portuguese. On these classifications,We use the lexical resource SentiWordNet, translated automatically to portuguese, and compare it’s performance with SentiLex, a lexical resource in portuguese. The analyses texts are comments are extracted from the website TripAdvisor, an online travel portal that provides informations and opinions related to tourism. The results show that the classification using the translated SentiWordNet has, in most of the cases, a better performance.