Cyberbullying Detection Using Msda

Abstract

Cyber bullying is any form of bullying which takes place online or through smart phones and tablets. The outcomes for victims under Cyberbullying may even be tragic such as the occurrence of self-injurious behaviour or suicides[1].One way is to provide a facility like “report abuse” to report abusive messages and behaviourisms in cyber bullying detection. But it focuses on assumption which reduces performance. And so we investigate one deep learning method named Stacked Denoising Auto encoder (SDA). SDA stacks several denoising auto encoders and concatenates the output of each layer as the learned representation. Each denoising auto encoder in SDA is trained to recover the input data from a corrupted version of it. The input is corrupted by randomly setting some of the input to zero. The denoising process helps the auto encoders to learn robust representation. We develop a new text representation model based on a variant of SDA: Marginalized Stacked Denoising Auto encoders (mSDA). An automatic extraction of bullying words based on word embeddings is proposed so that the involved human labour can be reduced. During training of smSDA, we attempt to reconstruct bullying features from other normal words by discovering the latent structure, i.e. correlation, between bullying and normal words. The intuition behind this idea is that some bullying messages do not contain bullying words.