Abstract
The number of online reviews has grown exponentially over the last years. As a result, several Sentiment Analysis Methods (SAMs) have been developed in order to extract automatically sentiments from text. In this work, we study polarity coherencies between reviewers and SAMs. To do so, we compare the polarity of the document evaluated by the user and the aggregated sentence polarity evaluated by three SAMs. The main contribution of this work is to show the flimsiness of user ratings as a generalization of the overall review sentiment.