改进回声状态网络的热点话题预测模型

为了提高网络热点话题的预测精度，针对传统回声状态网络存在的不足，提出一种改进回声状态网络的网络热点话题预测模型。首先将一维的网络热点话题时间序列重构成多维时间序列，然后采用改进回声状态网络对多维时间序列进行学习，建立网络热点话题预测模型，最后对模型性能进行仿真测试。结果表明，改进回声状态网络可对网络热点话题的变化趋势进行准确刻画，网络热点话题的预测精度得以提高，而具有更好的应用价值。

Hot topic prediction model based on improved echo state network

In order to improve the prediction precision of network hot topic and overcome the defect of traditional of echo state network, this paper proposes a novel hot topic prediction model based on improved echo state network. Firstly, one-dimension time series of network hot topic are reconstructed to multi-dimension, and then multi-dimension time series of network hot topic are input into improved echo state network to establish network hotspot topic prediction model, finally, the performance of model is tested by simulation experiments. The results show that the proposed model can accurately describe the hot topic change trend and improve the prediction precision of network hotspot topic and has good application values.