Cyber Bullying Detection Using Social and Textual Analysis

Huang, Qianjia; Singh, Vivek K.; Atrey, Pradeep K.

doi:10.1145/2661126.2661133

Cited by 171 publications

(95 citation statements)

References 5 publications

(10 reference statements)

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Squicciarini et al (2015) used personal, social network and content-specific features with a C4.5 Decision Tree classifier to detect bullies on online social networks such as MySpace and spring.me, and devised a set of rules to determine if a user's cyberbullying behaviour is instigated by the actions of another bully. Similarly, Huang et al (2014) found that including social features mined from a user's ego networks as input features to J48, Naïve Bayes, SVM, and ZeroR classifiers improved cyberbullying detection over the use of textual features alone. To detect bullying content in their sample of 1000 emails, Burn-Thorton and Burman (2012) found, however, that clustering using a kNN algorithm was sufficient.…”

Section: Supervised Learning Approachesmentioning

confidence: 92%

“…Serra and Venter (2011) is the earliest study in our sample using network-based features; they used total time present online using a mobile phone as a feature in their detection method. Nahar et al (2012), Huang et al (2014), and NaliniPriya and Asswini (2015) used ego networks as features to improve detection. NaliniPriya and Asswini (2015) used the ego network to compute temporal changes in the relationships between users, and uses the detected changes within the detection process.…”

Section: Network-based Featuresmentioning

confidence: 99%

“…NaliniPriya and Asswini (2015) used the ego network to compute temporal changes in the relationships between users, and uses the detected changes within the detection process. Huang et al (2014) discovered that the risk of cyberbullying is decreased in ego networks with many people and high interconnectivity (probably because in such networks there is likely to be increased social support for potential victims) but that a higher number of messages exchanged between users indicate a higher likelihood for cyberbullying. In Squicciarini et al (2015), the authors used the elapsed time between comments to measure the influence of cyberbullies on other users and the proliferation of bullying across a social network.…”

Section: Network-based Featuresmentioning

confidence: 99%

See 2 more Smart Citations

Approaches to Automated Detection of Cyberbullying: A Survey

Salawu

Lumsden

2020

IEEE Trans. Affective Comput.

139

View full text Add to dashboard Cite

Abstract-Research into cyberbullying detection has increased in recent years, due in part to the proliferation of cyberbullying across social media and its detrimental effect on young people. A growing body of work is emerging on automated approaches to cyberbullying detection. These approaches utilise machine learning and natural language processing techniques to identify the characteristics of a cyberbullying exchange and automatically detect cyberbullying by matching textual data to the identified traits. In this paper, we present a systematic review of published research (as identified via Scopus, ACM and IEEE Xplore bibliographic databases) on cyberbullying detection approaches. On the basis of our extensive literature review, we categorise existing approaches into 4 main classes, namely supervised learning, lexicon-based, rule-based, and mixed-initiative approaches. Supervised learning-based approaches typically use classifiers such as SVM and Naïve Bayes to develop predictive models for cyberbullying detection. Lexicon-based systems utilise word lists and use the presence of words within the lists to detect cyberbullying. Rule-based approaches match text to predefined rules to identify bullying, and mixed-initiatives approaches combine human-based reasoning with one or more of the aforementioned approaches. We found lack of labelled datasets and non-holistic consideration of cyberbullying by researchers when developing detection systems are two key challenges facing cyberbullying detection research. This paper essentially maps out the state-of-the-art in cyberbullying detection research and serves as a resource for researchers to determine where to best direct their future research efforts in this field.

show abstract

Section: Supervised Learning Approachesmentioning

confidence: 92%

Section: Network-based Featuresmentioning

confidence: 99%

Section: Network-based Featuresmentioning

confidence: 99%

See 1 more Smart Citation

Approaches to Automated Detection of Cyberbullying: A Survey

Salawu

Lumsden

2020

IEEE Trans. Affective Comput.

139

View full text Add to dashboard Cite

show abstract

“…From this perspective, the task of cyberbullying detection was previously approached as a classification task (Yin et al 2009) that involves data acquisition and pre-processing, feature extraction, and classification. These techniques were used mostly in targeting explicit textual cyberbullying language and rely on detecting features such as profanities (Yin et al 2009;Dinakar et al 2012;Dadvar et al 2013;Al-garadi et al 2016), bad words (Reynolds et al 2011;Huang et al 2014), foul terms (Nahar et al 2013), bullying terms (Kontostathis et al 2013;Nandhini and Sheeba 2015), pejoratives and obscenities (Chen et al 2012), emotemes and vulgarities (Ptaszynski et al 2010;Ptaszynski et al 2016), curses (Chatzakou et al 2017) or negative words (Van Hee et al 2015).…”

Section: Related Workmentioning

confidence: 99%

Detecting Discourse-Independent Negated Forms of Public Textual Cyberbullying

Power¹,

Keane²,

Nolan³

et al. 2018

J. Comp. Assist. Linguist. Res.

View full text Add to dashboard Cite

Cyberbullying is a risk associated with the online safety of young people and, in this paper, we address one of its most common implicit forms -negation-based forms. We first describe the role of negation in public textual cyberbullying interaction and identify the cyberbullying constructions that characterise these forms. We then formulate the overall detection mechanism which captures the three necessary and sufficient elements of public textual cyberbullying -the personal marker, the dysphemistic element, and the link between them. Finally, we design rules to detect both overt and covert negation-based forms, and measure their effectiveness using a development dataset, as well as a novel test dataset, across several metrics: accuracy, precision, recall, and the F1-measure. The results indicate that the rules we designed closely resemble the performance of human annotators across all measures.

show abstract

“…For feature selection, we aimed to determine the most effective features that were derived from multiple sources, as previous studies [15] suggested that classifiers that are based on a multisource of features can give better classification performance. For parameter tuning we aimed to determine the best setting for the number of trees, the size of leaf nodes and the depth of the trees in the RF classifier.…”

Section: Introductionmentioning

confidence: 99%

Using supervised machine learning algorithms to detect suspicious URLs in online social networks

Al-Janabi

Quincey

András

2017

Proceedings of the 2017 IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining 2017

View full text Add to dashboard Cite

Abstract-The increasing volume of malicious content in social networks requires automated methods to detect and eliminate such content. This paper describes a supervised machine learning classification model that has been built to detect the distribution of malicious content in online social networks (ONSs). Multisource features have been used to detect social network posts that contain malicious Uniform Resource Locators (URLs). These URLs could direct users to websites that contain malicious content, drive-by download attacks, phishing, spam, and scams. For the data collection stage, the Twitter streaming application programming interface (API) was used and VirusTotal was used for labelling the dataset. A random forest classification model was used with a combination of features derived from a range of sources. The random forest model without any tuning and feature selection produced a recall value of 0.89. After further investigation and applying parameter tuning and feature selection methods, however, we were able to improve the classifier performance to 0.92 in recall.

show abstract

Cyber Bullying Detection Using Social and Textual Analysis

Cited by 171 publications

References 5 publications

Approaches to Automated Detection of Cyberbullying: A Survey

Approaches to Automated Detection of Cyberbullying: A Survey

Detecting Discourse-Independent Negated Forms of Public Textual Cyberbullying

Using supervised machine learning algorithms to detect suspicious URLs in online social networks

Contact Info

Product

Resources

About