孟加拉语网络欺凌检测:用于高级分析的综合数据集
Gazi Tahsina Sharmin Jahin1, Firoze Maliha1, Nuren Nafisa1
1Department of Computer Science and Engineering, International Islamic University Chittagong, Kumira, IIUC Avenue, Sonaichhari, Chattogram, 4318, Chattogram, Bangladesh.
Data in brief
|January 1, 2026
概括
这项研究分析了孟加拉人在社交媒体上的网络欺凌行为,确定了关键主题,并通过mBERT和混合CNN方法等先进模型实现了93%的准确性. 局部可解释的模型不可知解释增强了模型的透明度.
科学领域:
- 计算语言学 计算语言学
- 社交计算社会计算
- 网络心理学 网络心理学
背景情况:
- 网络欺凌是一个重要的数字问题,由社交媒体放大.
- 对网络欺凌的研究广泛,但对孟加拉网络欺凌的研究有限.
- 这篇论文解决了孟加拉网络欺凌研究的稀缺性.
研究的目的:
- 分析孟加拉人在社交媒体平台上的网络欺凌行为.
- 识别负面孟加拉语评论的普遍主题和特征.
- 评估和比较各种机器学习模型的性能,以检测网络欺凌.
主要方法:
- 分析了超过7万个孟加拉人的社交媒体评论.
- 情绪分析以将评论分类为正面或负面.
- 潜在的迪里克莱特分配 (LDA) 用于负面评论的主题建模.
- 机器学习模型的应用和比较:支持矢量机器,XGBoost,CNN+BiLSTM+GRU,MBERT,XLM-R.
- 将BERT嵌入式与CNN和人工神经网络 (ANN) 模型集成.
- 使用局部可解释的模型不可知解释 (LIME) 来实现模型可解释性.
主要成果:
- mBERT实现了92%的最高准确率.
- 在CNN+BiLSTM+GRU混合模型达到91%的准确性.
- 将BERT嵌入到CNN和ANN模型中提高了性能,达到93%的准确性.
- 从负面评论中,LDA成功地提取了与年龄,性别,种族,宗教和各种类别相关的特征.
结论:
- 先进的深度学习模型,特别是mBERT和混合架构,对于检测孟加拉网络欺凌是有效的.
- 嵌入BERT显著提高了网络欺凌检测模型的性能.
- LIME为模型预测提供了有价值的见解,提高了对网络欺凌检测系统的透明度和信任.
- 这项研究为了解和打击孟加拉网络欺凌提供了一个强大的框架.
相关概念视频
02:04Bullying
8.8K
A modern form of aggression is bullying. As you learn in your study of child development, socializing and playing with other children is beneficial for children’s psychological development. However, as you may have experienced as a child, not all play behavior has positive outcomes. Some children are aggressive and want to play roughly. Other children are selfish and do not want to share toys. One form of negative social interactions among children that has become a national concern is...
8.8K
01:13Mass Analyzers: Overview
1.5K
The mass analyzer is a crucial component of the mass spectrometer. In the ionization chamber, the vaporized sample is bombarded with a high-energy electron beam to generate a radical cation and further fragment into neutral molecules, radicals, and cations. A series of negatively charged accelerator plates accelerate the cations into the mass analyzer. The mass analyzer separates ions according to their mass-to-charge (m/z) ratios and then directs them to the detector. The common types of mass...
1.5K
01:19Mass Analyzers: Common Types
1.3K
The quadrupole mass analyzer consists of four cylindrical metal rods arranged in a diamond carrying a DC voltage and a radio-frequency AC voltage. The motion of ions through the quadrupole depends on the field strength, causing only ions of a certain m/z to resonate successfully and strike the detector at a given field strength. Though the transmission rate for these analyzers is high, the exact elemental composition of the sample is not determined because of low resolution; however, they are...
1.3K