Machine Learning Techniques to Determine the Polarity of Messages on Social Networks

With the origin of Web 2.0, the Internet contains large amounts of user-generated information on an unlimited number of topics. Many entities such as corporations or political groups seek to gain knowledge through the opinions expressed by users. Social platforms such as Facebook or Twitter have proven to be successful for these tasks, due to the high volume of real-time messages generated and the large number of users that use them every day (Leis et al., J Med Internet Res 21(6):e14199, 2019, [1]). This paper focuses on the problem of Global Sentiment Analysis. Using the texts that compose the corpus in Spanish built by Sixto et al. (International conference on applications of natural language to information systems. Springer, Cham, 2016 [2]), and selecting the three most used classifiers by the state of the art of Naive Bayes, MSV and J48 through the Weka software.