Analyzing Labeled Cyberbullying Incidents on the Instagram Social Network

Abstract

Cyberbullying is a growing problem affecting more than half of all American teens. The main goal of this paper is to investigate fundamentally new approaches to understand and automatically detect and predict incidents of cyberbullying in Instagram, a media-based mobile social network. In this work, we have collected a sample data set consisting of Instagram images and their associated comments. We then designed a labeling study and employed human contributors at the crowd-sourced CrowdFlower website to label these media sessions for cyberbullying. A detailed analysis of the labeled data is then presented, including a study of relationships between cyberbullying and a host of features such as cyberaggression, profanity, social graph features, temporal commenting behavior, linguistic content, and image content. Using the labeled data, we further design and evaluate the performance of classifiers to automatically detect and predict incidents of cyberbullying and cyberaggression.

A Survey on Hate Speech Detection using Natural Language Processing

“…Contextual factors play an important role. For example, Hosseinmardi et al (2015) find that 48% of media sessions in their data collection were not deemed hate speech by a majority of annotators, even though they reportedly contained a high percentage of profanity words.…”

“…Moreover, there are certain kinds of meta-information for which conflicting results have been reported. For instance, Hosseinmardi et al (2015) report a correlation between the number of associated comments to a post and hate speech while Zhong et al (2016) report the opposite. (Both papers use Instagram as a source.)…”

A Novel Position-based Sentiment Classification Algorithm for Facebook Comments

“…They obtained a recall of 79% and a precision of 71% from text comments. For non-text features, the recall was slightly lower at 76% and the precision at 62% [14].…”

Harnessing the Power of Text Mining for the Detection of Abusive Content in Social Media

“…In order to avoid misspelling and abbreviation, Sood et al [28] improved on this static keyword-based approach by using the Levenshtein Distance. However, a high percentage of profanity words do not in fact constitute inappropriate content, so are not suitable for discrimination [16].…”

“…Wellknown classifiers have been used in this domain including Support Vector Machines (SVM) [9,16], Näive Bayes [4,13], logistic regression [29], and decision trees [25,16]. Classifier ensemble solutions have also proven successful.…”