Cyberbullying Detection in Social Networks Using Deep Learning Based Models

Dadvar, Maral; Eckert, Kai

doi:10.1007/978-3-030-59065-9_20

Cited by 48 publications

(34 citation statements)

References 16 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The experiments of the proposed model are conducted using Intel core i7@4GHz with 16 GB of RAM running on windows 10 platform and implemented using Python v2.7. The following eight different machine learning algorithms have been utilized to discover an optimal combination of base-classifiers in the initial stage and the meta-classifier in the second one: SVM, 38 NB, 28 DNN, 36 CNN, 34 RF, 39 AdaBoost, 37 LR, 39 and Bagging. 26 Table 2 displays the parameter settings of these classification algorithms and Table 3 showcases the empirically adjusted base-classifiers' hyper-parameter settings for different datasets.…”

Section: Resultsmentioning

confidence: 99%

“…Proposed model beat the best-known results for the datasets procured from Wikipedia, Twitter, and Formspring. Dadvar et al 36 proposed a Deep Neural Network (DNN)-based model for cyberbullying detection in social media platforms and extended their work to examine the transferability and adaptability of the model by using new dataset from YouTube social media. Rafiq et al 37 designed a multi-stage cyberbullying detection system that lessens the classification time dramatically and highly responsive in raising alerts without sacrificing accuracy.…”

Section: Related Workmentioning

confidence: 99%

“…The proposed model is compared with the state of the art cyberbullying detection techniques (ie, SVM, 38 Naïve Bayes (NB), 28 DNN, 36 Convolutional Neural Network (CNN), 34 Random Forest (RF), 39 AdaBoost, 37 Logistic Regression (LR), 39 and Bagging 26 ) based on machine learning classification approaches as stated in Table 1. It is depicted from the table that the literature work exploited a particular classifier, but the works did not combine their results that may result in improved performance.…”

Section: Related Workmentioning

confidence: 99%

See 2 more Smart Citations

Cuckoo inspired stacking ensemble framework for content‐based cybercrime detection in online social networks

Singh

Kaur

2020

Trans Emerging Tel Tech

View full text Add to dashboard Cite

In recent times, the topic of content-based cybercrime has gained significant attention. It is the requirement of today's era for social media companies to meet the challenges faced in differentiating the oppressive content in both a precise and proficient manner, thereby securing their clients. This article proposes a novel cuckoo inspired stacking ensemble framework that is the integration of Cuckoo Search and several machine learning models. The proposed framework automatically seeks for near-optimal combinations of classification techniques along with their tuning parameters for efficiently solving the problem of content-based cybercrime detection. Four datasets obtained from Formspring, ASKfm, and Twitter are used for testing purposes. The experimental results showcased significant improvement in the performance of classification on all the datasets in comparison to state-of-art classification models. The success rate of the proposed model with the excellent recall is 0.984 via 10-fold cross-validation demonstrates its high efficiency and effectiveness. INTRODUCTIONSocial media has evolved into a prestigious platform where people can share their emotions and has the advantage of being able to hide and perpetrate acts of violence against peers such as dating abuse, gang-related crimes, stacking, and bullying. Social media has also been used as a self-harming agent, especially cyber-suicide. 1 Young people of the 21st century cannot live without social media and the Internet-for example, Twitter, Facebook, and so on. Over 90% of young people use the Internet every day, and about 70% have active accounts on one of the social networking sites. 2,3 Verbal and face-to-face abuses are still more common than virtual abuse. Research indicates that most adolescents and children show little to no participation in abuses across social media networks (65-91%). 4,5 However, violence in the virtual world is a growing problem that requires additional research and prevention. Patchin and Hinduja 6 research shows that violence in social media and cyberbullying is becoming more and more common. The study included a random sample of 4441 teenagers aged 10-18 years from 37 schools. It was found that nearly 20% of teenagers in 2010 claimed to have been victims of cyberbullying, and 20% said that at one point in their lives they experienced bullying in cyberspace. The number of cybercrimes that have been developed and implemented on social media platforms such as Facebook is increasing. Communication on social networking sites is characterized by a higher degree of anonymity, which gives honest opportunities to increase hostility in interpersonal interactions. 7 The account of menacing, demoralization, teen pornography, University Grants Commission [No. F.159(JUNE 2014)/2014(NET)].

show abstract

Section: Resultsmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Cuckoo inspired stacking ensemble framework for content‐based cybercrime detection in online social networks

Singh

Kaur

2020

Trans Emerging Tel Tech

View full text Add to dashboard Cite

show abstract

“…Various heterogeneous graph neural networks are often used to understand the social characters of the users, such as the users' emotions, personality types, and interests [37]. Dadvar et al [38,39] performed a reproducibility study with deep learning based models that are conducted in [29], their reproduced experiment showed that deep learning based models with word embedding outperform the machine learning models, they also thought the effect of oversampling method should be assessed.…”

Section: Cyberbullying Detectionmentioning

confidence: 99%

Cyberbullying Detection in Social Networks Using Bi-GRU with Self-Attention Mechanism

et al. 2021

View full text Add to dashboard Cite

With the propagation of cyberbullying in social networks as a trending subject, cyberbullying detection has become a social problem that researchers are concerned about. Developing intelligent models and systems helps detect cyberbullying automatically. This work focuses on text-based cyberbullying detection because it is the commonly used information carrier in social networks and is the widely used feature in this regard studies. Motivated by the documented success of neural networks, we propose a complete model combining the bidirectional gated recurrent unit (Bi-GRU) and the self-attention mechanism. In detail, we introduce the design of a GRU cell and Bi-GRU’s advantage for learning the underlying relationships between words from both directions. Besides, we present the design of the self-attention mechanism and the benefit of this joining for achieving a greater performance of cyberbullying classification tasks. The proposed model could address the limitation of the vanishing and exploding gradient problems. We avoid using oversampling or downsampling on experimental data which could result in the overestimation of evaluation. We conduct a comparative assessment on two commonly used datasets, and the results show that our proposed method outperformed baselines in all evaluation metrics.

show abstract

“…Another direction focuses on the testing generalization of the current hatespeech classifiers (Agrawal and Awekar, 2018;Dadvar and Eckert, 2018;Gröndahl et al, 2018), where those methods are tested in other datasets and domains such as Twitter data (Waseem and Hovy, 2016), Wikipedia data (Wulczyn et al, 2017), Formspring data (Reynolds et al, 2011), and YouTube comment data (Dadvar et al, 2014).…”

Section: Related Workmentioning

confidence: 99%

HABERTOR: An Efficient and Effective Deep Hatespeech Detector

Tran¹,

Hu²,

Hu³

et al. 2020

Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP)

View full text Add to dashboard Cite

We present our HABERTOR model for detecting hatespeech in large scale user-generated content. Inspired by the recent success of the BERT model, we propose several modifications to BERT to enhance the performance on the downstream hatespeech classification task. HABERTOR inherits BERT's architecture, but is different in four aspects: (i) it generates its own vocabularies and is pre-trained from the scratch using the largest scale hatespeech dataset; (ii) it consists of Quaternionbased factorized components, resulting in a much smaller number of parameters, faster training and inferencing, as well as less memory usage; (iii) it uses our proposed multisource ensemble heads with a pooling layer for separate input sources, to further enhance its effectiveness; and (iv) it uses a regularized adversarial training with our proposed finegrained and adaptive noise magnitude to enhance its robustness. Through experiments on the large-scale real-world hatespeech dataset with 1.4M annotated comments, we show that HABERTOR works better than 15 state-ofthe-art hatespeech detection methods, including fine-tuning Language Models. In particular, comparing with BERT, our HABERTOR is 4∼5 times faster in the training/inferencing phase, uses less than 1/3 of the memory, and has better performance, even though we pretrain it by using less than 1% of the number of words. Our generalizability analysis shows that HABERTOR transfers well to other unseen hatespeech datasets and is a more efficient and effective alternative to BERT for the hatespeech classification.

show abstract

Cyberbullying Detection in Social Networks Using Deep Learning Based Models

Cited by 48 publications

References 16 publications

Cuckoo inspired stacking ensemble framework for content‐based cybercrime detection in online social networks

Cuckoo inspired stacking ensemble framework for content‐based cybercrime detection in online social networks

Cyberbullying Detection in Social Networks Using Bi-GRU with Self-Attention Mechanism

HABERTOR: An Efficient and Effective Deep Hatespeech Detector

Contact Info

Product

Resources

About