Cyberbullying poses a digital threat to society. In this survey, we explain what cyberbullying is and its various forms. We focus on social media platforms and instant messaging apps that are susceptible to cyberbullying, discussing how we can identify such behavior in these spaces. Moving on, we conduct a systematic review of publicly available datasets in different languages, exploring techniques for data preprocessing, feature representation, and methodologies used in textual analysis for cyberbullying detection. We specifically look at natural language-based and platform-specific preprocessing methods. We also cover popular feature representation techniques like sentiment analysis, user information, text summarization, symbols, images, and word embedding for detecting cyberbullying. Next, we categorize existing techniques, including machine learning and neural networks, highlighting research gaps. Additionally, we discuss the challenges associated with current datasets and methods. This survey aims to provide early researchers with insights into cyberbullying literature and guide them in exploring potential research directions.

中文翻译：

---

防范网络伤害：防止网络欺凌的文本分析调查

网络欺凌对社会构成数字威胁。在这项调查中，我们解释了什么是网络欺凌及其各种形式。我们重点关注容易受到网络欺凌的社交媒体平台和即时通讯应用程序，讨论如何识别这些空间中的此类行为。接下来，我们对不同语言的公开数据集进行系统审查，探索数据预处理技术、特征表示以及网络欺凌检测文本分析中使用的方法。我们特别关注基于自然语言和特定于平台的预处理方法。我们还涵盖了流行的特征表示技术，例如情感分析、用户信息、文本摘要、符号、图像和用于检测网络欺凌的词嵌入。接下来，我们对现有技术（包括机器学习和神经网络）进行分类，突出研究差距。此外，我们还讨论了与当前数据集和方法相关的挑战。这项调查旨在为早期研究人员提供对网络欺凌文献的见解，并指导他们探索潜在的研究方向。