基于熵值法的网络舆情组合预测研究

Research on Network Public Opinion Combination Forecasting Based on Entropy Method

【目的/意义】互联网成为网民情感宣泄的平台使得网络舆情影响力剧增,构建模型对网络舆情的发展进行预测具有现实意义。【方法/过程】针对网络舆情趋势预测及时性的需求,以事件标签确定待选历史数据,通过模糊理论的模糊逻辑构建模糊时间序列预测模型,同时构建BP神经网络预测模型,以组合预测的方式提高整体的预测精度。【结果/结论】通过实验分析结果表明,预测模型可以在一定程度将预测的时间区间前置,实现"早期"预测。

【Purpose/significance】The internet has become a platform for emotional catharsis which makes the influence of network public opinion surge. Therefore, it is of practical significance to construct the model to predict the development of public opinion.【Method/process】For the requirement of timely prediction of network public opinion, the paper uses the event tag to determine the historical data. The fuzzy time series prediction model is constructed by the fuzzy logic of fuzzy theory. At the same time, BP neural network prediction model is constructed to improve the overall prediction accuracy by combining forecasting.【Result/conclusion】The experimental results show that the forecasting model can advance the prediction interval to a certain extent to achieve the ＇early＇ prediction.