A Sementic Enhanced Technique For Cyberbulling Detection

Abstract

Now-a- days internet is mostly useful for the people for school, work, and social use, so too do more people turn to the Internet to take out their frustrations and aggression. One form of cyber aggression has been gaining the attention of both researchers and the public in recent years: cyber bullying. Cyber bullying is typically defined as aggression that is intentionally and repeatedly carried out in an electronic context (e.g., e-mail, blogs, instant messages, text messages) against a person who cannot easily defend him-or herself. Many researchers have noted that cyber bullying is occurring at widespread rates among youth and adults, with some studies showing nearly 75% of school-age children experiencing this form of aggression at least once in the last year. The experience of cyber bullying has been linked with a host of negative outcomes for both individuals and organizations (e.g., schools), including anxiety, depression, substance abuse, difficulty sleeping, increased physical symptoms, decreased performance in school, absenteeism and truancy, dropping out of school. To deal with these problems, In this paper, we investigate one deep learning method named stacked denoising auto encoder (SDA). We develop a new text representation model based on a variant of SDA: marginalized stacked denoising auto encoders (mSDA), which adopts linear instead of nonlinear projection to accelerate training and marginalizes infinite noise distribution in order to learn more robust representations. Our proposed Semantic-enhanced Marginalized Stacked Denoising Auto encoder is able to learn robust features from BoW representation in an efficient and effective way. These robust features are learned by reconstructing original input from corrupted (i.e., missing) ones. The new feature space can improve the performance of cyber bullying detection even with a small labeled training corpus.