【摘要】 在全媒体时代，重大突发事件网络舆情数据规模大、传播范围广、群体交互性强，短时间内可能造成舆情态势的迅速恶性演化，对社会稳定造成严重威胁。因此，实现对重大突发事件网络舆情趋势的准确预测，有利于降低事件“衍生灾害”发生的风险。论文提出一种基于因子分析法（FA）与时序卷积神经网络（TCN）的组合预测模型FA-TCN，通过考虑引发舆情危机的影响因素建立指标体系，利用因子分析法进行舆情数据降维。针对网络舆情的演化趋势进行实例分析，结果表明论文提出的FA-TCN预测模型的各项指标均优于对比模型，更好地拟合了网络舆情的实际发展趋势，为相关部门准确把控舆情态势并制定科学高效的应急方案提供了理论支撑。更多还原

【Abstract】 In the era of omni-media,network public opinion on major emergencies has the characteristics of large data scale,wide dissemination range,and strong group interaction,which may cause rapid and vicious evolution of public opinion in a short period of time,posing a serious threat to social stability. Therefore,accurate prediction of the trend of network public opinion for major emergencies is conducive to reducing the risk of "derived disasters". This paper proposes a combined forecasting model FA-TCN based on factor analysis(FA)and time series convolutional neural network(TCN). By considering the influencing factors of public opinion crisis,an index system is established,and the factor analysis method is used to reduce the dimension of public opinion data. An example analysis is carried out on the evolution trend of network public opinion,and the results show that the FA-TCN prediction model proposed in this paper is better than the comparison model in every index. The model better fits the actual development trend of online public opinion,and provides theoretical support for relevant departments to accurately understand the situation of public opinion and formulate scientific and efficient emergency plans.更多还原