本研究试图通过文本分析的方法，使用文本对网络暴力者的暗黑三人格进行预测， 并比较网络暴力者与非网络暴力者的暗黑三人格差异。采用简式暗黑三人格量表对 212 名参与者进行测查，并获得有效用户的知乎平台回答文本，使用文本分析得到的文本特征进行机器学习回归分析， 使用得到的回归模型来预测网络暴力者与非网络暴力者的暗黑三人格，并进行差异检验。结果显示暗黑三人格与文本特征之间存在显著的相关，网络暴力者与非网络暴力的文本特征之间存在显著差异， 网络暴力者在精神病态维度上显著高于非网络暴力者，在自恋与马基雅维利主义维度上稍高于非网络暴力者，但无显著差异。

The research tries to use text analytics to predict dark triad of cyber-aggressors, and compare dark triad of cyber-aggressors with that of non-cyber-aggressors. Test 212 participants with the Short Dark Triad and then crawl answers of whose data is valid from Zhihu. Use machine learning regression algorithm to analyze text characteristics extracting from the content, and predict dark triad of cyber-aggressors and non-cyber-aggressors, and test difference between them. The result shows that correlation between dark triad and text characteristics is significant, and text characteristics between cyber-aggressors and non-cyber-aggressors is significantly different. Psychopathy of cyber-aggressors is significantly higher than non-cyber-aggressors, while Narcissism and Machiavellianism of cyber-aggressors is little but not significantly higher than non-cyber-aggressors.