A Novel Approach for Cyberbullying on Social Media Using SEMAE
A. Lahari, A. Ganesh · 2017
As a symptom of progressively mainstream online networking, cyberbullying has risen as a difficult issue tormenting children, adolescents and youthful grown-ups.Machine learning strategies make programmed discovery of tormenting messages in online networking conceivable, and this could construct a strong and safe electronic interpersonal interaction condition.In this huge research zone, one essential issue is capable and discriminative numerical depiction learning of texts.In this paper, we propose another depiction learning method to deal with this issue.Our system named Semantic-Enhanced Marginalized Denoising Auto-Encoder (smSDA) is created by methods for semantic development of the conspicuous significant learning model stacked denoising auto encoder.The semantic growth contains semantic dropout tumult and sparsity objectives, where the semantic dropout clatter is arranged in light of region learning and the word embeddings methodology.Our proposed system can mishandle the covered highlight structure of tormenting information and take in a fiery and discriminative depiction of substance.