Sentiment Informed Cyberbullying Detection in Social Media

Abstract

Cyberbullying is a phenomenon which negatively affects individuals. Victims of the cyberbullying suffer from a range of mental issues, ranging from depression to low selfesteem. Due to the advent of the social media platforms, cyberbullying is becoming more and more prevalent. Traditional mechanisms to fight against cyberbullying include use of standards and guidelines, human moderators, use of blacklists based on profane words, and regular expressions to manually detect cyberbullying. However, these mechanisms fall short in social media and do not scale well. Users in social media use intentional evasive expressions like, obfuscation of abusive words, which necessitates the development of a sophisticated learning framework to automatically detect new cyberbullying behaviors. Cyberbullying detection in social media is a challenging task due to short, noisy and unstructured content and intentional obfuscation of the abusive words or phrases by social media users. Motivated by sociological and psychological findings on bullying behavior and its correlation with emotions, we propose to leverage the sentiment information to accurately detect cyberbullying behavior in social media by proposing an effective optimization framework. Experimental results on two realworld social media datasets show the superiority of the proposed framework. Further studies validate the effectiveness of leveraging sentiment information for cyberbullying detection.

Subversive Toxicity Detection using Sentiment Information

“…They found its inclusion helped improve results, but that a more sophisticated source of information than simple keyword detection was required. And the study of (Dani et al, 2017) used the sentiment of messages, as measured by the SentiStrength online system, as one of several features to detect cyberbullying messages. However, an in-depth analysis of how sentiment can benefit toxicity detection has not been done in any of these papers, and a study of the use of sentiment in a subversive context has never been done.…”

Unsupervised Cyberbullying Detection via Time-Informed Gaussian Mixture Model

“…Dinakar et al [13] concatenated TF-IDF scores, POS tags of frequent bigrams, and profane words as content features to detect cyberbullying on a manuallylabeled corpus of YouTube comments. Dani et al [9] sought to incorporate sentiment into the content features by capturing the sentiment consistency of bullying and non-bullying posts. Most recently, Ziems et al [49] characterized cyberbullying using five explicit factors to represent its social and linguistic aspects.…”

Cyberbullying Detection Through Machine Learning: Can Technology Help to Prevent Internet Bullying?

“…Therefore, ML algorithms could serve to early and automatically detect cyberbullying, and to foster intervention protocols' timely activation. In the last decade, researchers have tested a variety of ML techniques such as victims' sentiment informed analysis, textual, and semantic analysis, and user features' analysis (e.g., gender) [46,47,58,49]. Plus, these methods allowed to detect a variety of cyberbullying outcomes including binary classifications (e.g., being or not-being involved), role identification in cyberbullying dynamics and the severity of consequences [50,51,52].…”