Automatic document categorization

• Main Author: Zhou, Anna
• Other Authors: Mao Kezhi
• Format: Final Year Project
• Language: English
• Published: 2016
• Subjects: DRNTU::Engineering
• Online Access: http://hdl.handle.net/10356/67886
• Institution: Nanyang Technological University

With the increasing popularity of social media network in the recent years, the concerns have been raised for the exposure of cyber bullying. The harmful information brings huge negative impact on the mental health of people who are exposed to them, especially teenagers. Therefore, it is essential to find an effective way of cyber bullying detection. In this paper, we proposed two different models for the text representation and feature extraction. Introduction to the topic and some related work were presented firstly for a better understanding of the topic. Then the concept of the two text representation models Embedding Enhanced Bag-of-Words model and Bullying-Word-Filter model were introduced. In the experiment part, we applied these two models with some manually labeled tweets and did the testing. The performances of prediction scores were illustrated. In the second part, with the classifiers trained in the first part, a case study concentrating on the cyber bullying cases in Singapore was done. It wasshown in the paper that our proposed models outperformed many existing models and worked efficiently in cyber bullying detection. In the future, more works are supposed to be finished.