Emotion-based classification through fuzzy entropy enhanced FCM clustering

The proposed work introduces an emotion-based classification method from a social media data stream, such as Twitter, to detect the main emotions triggered by a social trend. The case study is analyzed by exploiting an extension of the Fuzzy C-Means (FCM) clustering algorithm, that exploits the fuzzy entropy measure to assess the data distribution and overcome the FCM’s sensitivity to the random cluster initialization. The remainder of the paper is organized as follows. Section 11.2 introduces a literature overview about sentiment analysis, emotion extraction approaches, in the Twitter’s domain. Section 11.3 provides aan overview on the main emotion-based models from the literature. The theoretical background on FCM as well as the fuzzy entropy measure and the proposed entropy based FCM model are presented in Section 11.4, while Section 11.5 describes the approach for emotion detection from an Italian reference Twitter dataset. Finally, Section 11.6 compares the results achieved by the entropy-based FCM with the classical FCM algorithm, showing that the proposed method converges faster and provides promising classification performance, evaluated by the common metrics such as accuracy, precision, and F1-score. Section 11.7 outlines the concluding remarks