Data Mining for Cyberbullying and Harassment Detection in Arabic Texts

Bashir, Eman; Bouguessa, Mohamed

doi:10.5815/ijitcs.2021.05.04

Cited by 10 publications

(8 citation statements)

References 22 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The cross-entropy loss that was evaluated throughout various epochs within the configuration demonstrated effective convergence, suggesting an optimal level of performance for the model. In the results obtained by [20], it was shown that cyberbullying detection by LSTM achieved an accuracy of 72%, and this performance increased from 65% to 72% after 10 epochs, while machine learning classifiers only achieved an accuracy of less than 70%. Finally, according to [25] they evaluated the LSTM model and revealed that LSTM was a viable method for identifying comments that were aggressive online, with an accuracy of 93.84% and an f score of 0.94.…”

Section: Resultsmentioning

confidence: 97%

See 1 more Smart Citation

Deep Learning Algorithms for Cyber-Bulling Detection in Social Media Platforms

Obaida,

Elkaffas,

Guirguis

2024

IEEE Access

View full text Add to dashboard Cite

Social media platforms are among the most widely used means of communication. However, some individuals exploit these platforms for nefarious purposes, with "cyberbullying" being particularly prevalent. Cyberbullying, which involves using electronic means to harass or harm others, is especially common among young people. Consequently, this study aims to propose a model for detecting cyberbullying using a deep learning algorithm. Three datasets from Twitter, Instagram, and Facebook were utilized to predict instances of bullying using the Long Short-Term Memory (LSTM) method. The results obtained revealed the development of an effective model for detecting cyberbullying, addressing challenges faced by previous cyberbullying detection techniques. The model achieved accuracies of approximately 96.64%, 94.49%, and 91.26% for the Twitter, Instagram, and Facebook datasets, respectively.

show abstract

Section: Resultsmentioning

confidence: 97%

“…This particular embedding option (Keras) stands out since it is task-specific. By including semantics instead of just features obtained from raw text, word embedding enhanced the suggested model's level of accuracy in comparison to other conventional detection approaches [20].…”

Section: Word Embeddingmentioning

confidence: 99%

Deep Learning Algorithms for Cyber-Bulling Detection in Social Media Platforms

Obaida,

Elkaffas,

Guirguis

2024

IEEE Access

View full text Add to dashboard Cite

show abstract

“…For example in the classification method, several classification techniques are compared, including K-Nearest Neighbor, Random Forest, Decision Tree, and Naive Bayes. Prediction of cyber bullying is done using the K-Nearest Neighbor technique but focuses on positive comments even though there are potential negative comments [10].…”

Section: Methodsmentioning

confidence: 99%

“…Data mining analysis was also proposed by Bashir with the concept of analyzing different Arabic texts previously in the Korean context. Arabic has a high complexity [10]. The concept that is built is only a protection not to predict.…”

Section: Theoretical Reviewmentioning

confidence: 99%

Cyberbullying Prediction As Cyber Counseling Tools With Data Mining Classification

Pamuji¹,

Setiawan

2022

Bit

View full text Add to dashboard Cite

Pertumbuhan data dan informasi semakin pesat sedangkan pengguna perangkat teknologi informasi terus meningkat. Apalagi kemudahan akses informasi didukung dengan hadirnya teknologi komunikasi bergerak yang dimiliki oleh hampir setiap pengguna. Saat ini terjadi peningkatan jumlah pengguna yang signifikan, yang dinilai berpeluang untuk hadirnya kejahatan dunia maya, terutama dalam kasus bullying. Masalah utama adalah sulitnya memprediksi potensi cyberbullying karena dunia maya penuh dengan anonimitas. Kasus cyberbullying termasuk kejahatan dunia maya di luar keamanan komputer jika dilihat dari perspektif perilaku. Oleh karena itu, cyberbullying menganalisis perilaku ketika melakukan tindakan negatif. Dalam studi ini, kami telah menyelidiki dan memprediksi kecenderungan bullying menggunakan pendekatan data mining. Analisis studi kasus yang disajikan dan teknis pelaksanaannya melalui pendekatan data mining. Menggunakan data mining sebagai alat, kami menggunakan beberapa metode klasifikasi dengan tujuan perbandingan dan metode mana yang terbaik untuk analisis. Metode klasifikasi dalam data mining adalah K-NN, Random Forest, Decision Tree, dan Naive Bayes. Dengan perbandingan dari empat metode, ada tiga kelas. Ada tiga kelas, yaitu tidak ada potensi bullying, kekerasan dan hinaan. Dengan demikian, tiga kelas terdeteksi berdasarkan hasil investigasi dan beberapa teknik dievaluasi dengan menggunakan perbandingan kinerja masing-masing teknik. Hasil akhir akan menunjukkan teknik decision tree berkinerja terbaik dikarenakan selama tahap preprocessing tidak membutuhkan waktu lama. tambahannya adalah karena kemampuan membreakdown setiap cabang dari data sehingga waktu yang dibutuhkan pada data cleaning lebih cepat dengan akselerasi 7%.

show abstract

“…The digital environment may open new ways to perpetrate violence against children. Open access to the Internet, difficulties in tracking the abuser, fear of parents' reaction to cyberthreats [2][3] and, as a result, the concealment of cyberbullying [4][5] make the child easy prey. Crises, such as pandemics, may lead to an increased risk of online harm, considering that children spend more time on virtual platforms [6].…”

Section: Introductionmentioning

confidence: 99%

State and Legal Policy for the Protection of Children in the Digital Environment

Hradova¹,

Zhytnyi²,

Tereshchuk³

2021

Proceedings of the International Conference on Social Science, Psychology and Legal Regulation (SPL 2021)

View full text Add to dashboard Cite

Digital technology and the Internet have made our lives much easier by providing opportunities to work, study, and shop online. Today half of humanity uses the Internet and social networks, which have brought risks and dangers despite significant benefits. In the digital environment the most vulnerable category of population are children, who may be victims of cyber-violence, especially cyberbullying or sexting According to the results of an anonymous survey among students, we were able to find out that 54% of respondents had faced cyberbullying. As for sexting, about 50% of interviwed had received an offer to send their intimate photo / video, but only 20% of them agreed. During our research we tried to find out the reasons for the growth of cyberbullying in the digital environment, as well as ways to counter online harassment. According to the respondents, criminalization of counteracting cyberbullying and sexting, as well as proper state policy should be the most effective method. It should include the development of relevant regulations, various campaigns to protect children from cyberbullying, the creation of hotlines for victims of cyberbullying and so on. The public authority which is responsible for developing of the children's digital rights and ensuring their safety in the digital environment is the Ministry of Digital Transformation. It has proposed the Concept for the Development of Digital Rights of Children, which provides with the implementation of many measures to protect children`s rights in the digital environment, including the criminalization of cybergrooming.

show abstract

Data Mining for Cyberbullying and Harassment Detection in Arabic Texts

Cited by 10 publications

References 22 publications

Deep Learning Algorithms for Cyber-Bulling Detection in Social Media Platforms

Deep Learning Algorithms for Cyber-Bulling Detection in Social Media Platforms

Cyberbullying Prediction As Cyber Counseling Tools With Data Mining Classification

State and Legal Policy for the Protection of Children in the Digital Environment

Contact Info

Product

Resources

About