Ameliorated feature construction mechanisms for emotions identification in social media texts

Abstract

The feature selection and feature extraction techniques proposed in this thesis are applicable to any text-based emotion detection applications including intention mining, sentiment analysis, depression analysis, etc. Due to the pervasive nature of depression (Bueno-Notivol et al. 2021), its severity, and the prospective of utilizing texts from social media forums to detect depression, a case study of depression diagnosis is investigated in this research. And the experiments are carried out using the depression detection datasets collected from Twitter and Reddit. newline newline newline