Scholarlink

<p>The growing usage of Social Media Platforms (SMP) like Twitter, Instagram, and Facebook has resulted in a significant surge in Cyberbullying (CB). Preventing CBon digital platforms is difficult because bullies use sarcasm and passive-aggressive tactics. Therefore, this research proposes a novel model of Enhancing Social Media Security through Deep Learning based Cyberbullying Detection Optimized with Snow Ablation Optimizer (ESM-DL-CBD). Initially, data is collected from Large CBDataset and Comprehensive Dataset for Automated CBDetection (CDACD) dataset. Then, the input data is fed into pre-processing for removing the noise, tokenization, punctuation, symbols removal and removal of stop words. After, pre- processed data is given to feature extraction using Fuzzy Bag-of-Words (FBoW) to extract bigram, unigram and trigram parameter features. Then Localized Sparse Incomplete Multi-view Clustering (LSMC) to automatically examine the text data and generates clusters of words. Then, the examined text data and created cluster words is fed to Contrastive Multi-Level Graph Neural Network (CMGNN) for detecting the CB as Bullying and Non Bullying. In general, CMGNN does not show some adaption of optimization approaches for finding ideal parameters to ensure accurate detection of the CB. Hence, Snow Ablation Optimizer (SAO) is utilized to optimize CMGNN for precisely detecting the CB. Experimental results verify that the proposed ESM-DL-CBD method has significant improvements over the existing approaches. The ESM-DL-CBD method has achieved relative accuracy improvements of 30.73%, 28.35%, and 29.62%, as well as precision improvements of 20.56%, 18.25%, and 31.60%. These findings demonstrate that the ESM-DL-CBD model significantly enhances social media security by enabling more accurate and efficient CB detection.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Enhancing social media security through deep learning based cyberbullying detection optimized with snow ablation optimizer

  • 期刊论文

  • 发表日期：

    2026年7月12日

查看全文
  • U. Srinadh,
  • P. Sankara Rao

摘要

The growing usage of Social Media Platforms (SMP) like Twitter, Instagram, and Facebook has resulted in a significant surge in Cyberbullying (CB). Preventing CBon digital platforms is difficult because bullies use sarcasm and passive-aggressive tactics. Therefore, this research proposes a novel model of Enhancing Social Media Security through Deep Learning based Cyberbullying Detection Optimized with Snow Ablation Optimizer (ESM-DL-CBD). Initially, data is collected from Large CBDataset and Comprehensive Dataset for Automated CBDetection (CDACD) dataset. Then, the input data is fed into pre-processing for removing the noise, tokenization, punctuation, symbols removal and removal of stop words. After, pre- processed data is given to feature extraction using Fuzzy Bag-of-Words (FBoW) to extract bigram, unigram and trigram parameter features. Then Localized Sparse Incomplete Multi-view Clustering (LSMC) to automatically examine the text data and generates clusters of words. Then, the examined text data and created cluster words is fed to Contrastive Multi-Level Graph Neural Network (CMGNN) for detecting the CB as Bullying and Non Bullying. In general, CMGNN does not show some adaption of optimization approaches for finding ideal parameters to ensure accurate detection of the CB. Hence, Snow Ablation Optimizer (SAO) is utilized to optimize CMGNN for precisely detecting the CB. Experimental results verify that the proposed ESM-DL-CBD method has significant improvements over the existing approaches. The ESM-DL-CBD method has achieved relative accuracy improvements of 30.73%, 28.35%, and 29.62%, as well as precision improvements of 20.56%, 18.25%, and 31.60%. These findings demonstrate that the ESM-DL-CBD model significantly enhances social media security by enabling more accurate and efficient CB detection.