Every day a large amount of data circulates on the Internet, because it is enormous to the quantities of computers and electronic equipment that have access to the internet, producing information in the form of images, videos and texts, with that, a way of analyzing and extracting these information. Therefore, this work presents concepts and techniques to find patterns of feelings and emotions, through the use of techniques and tools, especially in the area of text mining. The purpose of the procedures performed in this work was to develop a method to extract the emotional and sentimental profiles of the comments of the users of Last Day of June, where the results and information extracted from the analysis of feelings were presented. Three classification algorithms, Naive Bayes, Support Vector Machine (SVM) and K-Nearest Neighbors (KNN) were also used to predict the class of elements according to the feelings identified in the comment analysis step, between the SVM with the Radial kernel was the one that presented the best precision, with 79%, followed by the KNN with 3 nearest neighbors, with 75% and finally the Naive Bayes, with 62%.
|
