Avaliação do desempenho de métodos de análise de sentimentos na presença das figuras de linguagem sarcasmo e ironia

Abstract

Sentiment analysis is a study area expanding, applied to numerous environments (financial, political, academic, business or and communication), whose purpose is to search for messages posted on social media, and through these to identify and classify people’s opinions about particular item as positive or negative. Classify the sentiment expressed in opinionated messages is such an important task that currently companies invest a lot of money in collecting this type of information and the development of methods and techniques to classify the feeling that they express, to use the results as useful information in preparation marketing and sales strategies efficiently, or the academic sector to discover relevant factors society. Currently many of these methods began to be developed and implemented in various types of applications and tools that perform sentiment analysis on shared messages on social networks. However, many of these methods are being used in developing applications without a concrete understanding of its applicability in different contexts, its advantages, limitations and efficiency compared to other methods. Among these contexts, is one of the major problems facing the sentiment analisys, which is the difficulty of methods to properly analyze messages with content sarcastic or ironic, since these linguistic phenomena have the characteristic of transforming the polarity or meaning a positive or negative statement into its opposite. In this context the present study sought to evaluate and compare quantitatively the performance of different methods existing in the literature to classify the polarity messages with the presence of sarcasm.