Abstract:

The term weighting is an important step that directly affects the results in the text classification. However, the behavior of the weighting method, depending on different prediction techniques, can change in the emotional analysis, which is treated as a text classification problem. In this study, the methods recently recommended for different fields of study such as information-reference, text classification, document filtration have been applied in the Twitter emotional analysis and the impact on the results has been studied. Two different models were used, the word bag (BoW) and the character level N-gram, when the properties were removed. The experiments were applied on data sets consisting of Turkish and English Twitter messages. The emotional classification of Twitter messages is carried out with the theme model based on the Secret Dirichlet Appointment (LDA). In the classification phase, the support vector machine (SVM) algorithm was used. According to experimental findings, the most effective term emphasis method that can be used in the Twitter Emotional Analysis studies has been suggested.

Keywords: