网络暴力事件演化与预测研究

目的/意义]准确掌握网路暴力事件的演化路径,并及时预测潜在的网络暴力事件,为相关部门治理舆情提供参考。方法/过程]研究了网络暴力舆情事件的演进阶段、演进要素及演化路径;从舆情本体、舆情传播、舆情反应三个方面抽取网络暴力事件的相关特性。面向不平衡数据子集,基于多层感知机提出一种融合集成噪声识别与SMOTE算法的网络暴力预测模型。结果/结论]提出的预测模型准确率达88.7%,且具有较好的泛化能力。暴露隐私信息是网络暴力事件发生最关键的因素。

Research on the Evolution and Prediction of Cyber Violence Incidents

Purpose/significance]The paper is to accurately grasp the evolution path of cyber violence incidents,and predict potential cyber violence incidents in time,to provide references for relevant departments to govern online public opinions.Method/process]The paper studies the evolution stage,evolution elements and evolution path of online violent public opinion incidents;extracts the relevant characteristics of cyber violent incidents from three aspects:the main body of public opinion,the spread of public opinion and the response of public opinion.For a subset of unbalanced data,a cyber violence prediction model that integrates noise recognition and SMOTE algorithm is proposed based on a multilayer perceptron.Result/conclusion]The prediction model above mentioned has an accuracy rate of 88.7%and good generalization ability.Exposing private information is a key factor in the occurrence of cyber violence.