ABSTRACT

This chapter briefly discusses two essential aspects of the author, namely, the author's aggression and the author's profile. The identification of aggression is classified into three classes, overt, covert and non-aggressive. On the other hand, profiling identifies two properties, age and gender. The chapter also explains various machine learning concepts incorporated into deep neural networks. It also provides insight into the previous work done on author aggression and profile identification. For the purposes of this experiment and in-depth analysis, Twitter and Facebook datasets are used. All of the proposed architectures perform on average better than state-of-art models. For English Twitter, the aggression model achieves 60.09% accuracy and the profiling model can achieve 68.02% accuracy.