Date of Publication :2nd August 2017

Abstract: Social media now a day is a major communication bridge among peoples almost around the world which helps them to share their day to day activities to the one who is in contact with them in that network; hence it acts as an intermediary among them. As social media have many excellent features for better communication, it does have the many drawbacks which may hurt the intensions of the peoples sometimes due to some weird actions of the users on social media, such as teasing or posting something which hurts the user intensions by using bullying or tormenting words. The major criterion to deal this problem is learning robust representations of texts through text mining concepts in machine learning and NLP. So to prevent this kind of activities on social networks, we are proposing a technique named Semantic-Enhancement for Marginalized Denoising Auto- Encoder(smSDA), which is an extension of the admired machine learning technique Stacked Denoising Auto- Encoder (SDA).

Reference :