Review of Sentiment Analysis Techniques in Social Networks

Abstract

This article develops a review of the techniques used for sentiment analysis applied to messages on social networks. Sentiment analysis is a task from the field of Artificial Intelligence known as Natural Language Processing and seeks to detect the sentiment polarity expressed by a person in a short message or in a document. Currently there are three types of techniques from which the proposed sentiment analysis models are derived: Lexicons, traditional machine learning and deep learning – in addition to hybrid approaches that combine at least two of these techniques to improve classification performance. Sentiment analysis is a currently-valid task and its importance has grown along with the massification of social networks, which allow a massive generation of text that can be classified.