中文摘要：

网络热点话题具有时变性和非线性，灰色、负面热点话题对社会稳定产生不利影响。为了提高网络热点话题的预测精度，提出一种基于改进回声状态网络的热点话题预测模型（MESN）。首先构建网络热点话题的学习样本，然后采用回声状态网络建立网络热点话题预测模型，并利用改进粒子群优化算法对回声状态网络参数进行优化，建立最优网络热点话题预测模型，最后应用具体网络热点话题数据进行仿真实验。结果表明，该模型不仅提高了网络热点话题预测精度，而且加快了网络热点话题的建模速度，可以满足网络热点话题在线预测。

英文摘要：

Hot network topics are time varying and nonlinear, those grey and negative hot topics can produce adverse effects on social stability. In order to improve the prediction precision of hot network topics, in this paper we proposed a hot topics prediction model which is based on modified echo state network （MESN）. First, we constructed the learning samples of hot network topics; then used echo state network to build the prediction model of hot network topics, and the improved particle swarm optimisation to optimise the parameters of echo state network; finally, we put the specific hot network topics data into simulation experiment. Results showed that the proposed model improved the prediction accuracy of hot network topics, and speeded up modelling the hot network topics, it could satisfy the online prediction of hot network topics.