Abstract: Aiming at the small amount of current public opinion event,it was necessary to learn from various types of online public opinion event.Although the trend of the same type was similarities,it was difficult to fit the development trend of new public opinion events based on historical data alone.A multi-fractal optimization-based public opinion monitoring method was proposed.The multi-fractal dimension was used to analyze the fractal characteristics of the public opinion propagation situation.According to the public opinion incident data,the development trend of subsequent public opinion was judged.Based on the judgment results,the fractal interpolation theory was introduced to achieve a public situation prediction industry.The electric power enterprise was taken as an example to verify the feasibility of opinion propagation.The experimental results showed that the method realized real-time monitoring and effective control of the public opinion event,and assisted the modern manufacture to control and guide the benign development of network public opinion.

Key words: modern manufacturing, opinion monitoring, multi-fractal, fractal interpolation, situation prediction

摘要： 针对当前制造业转型舆情事件数据量不足,需借鉴各类网络舆情事件的丰富历史数据,同时虽然同一类型的舆情发展态势存在共性,但仅凭历史数据往往难以较好地拟合并评估新生舆情事件的发展态势。鉴于此,采用基于多重分形态势优化方法实现新生突发网络舆情事件监测。首先,使用多重分形维数分析舆情传播态势,并根据抓取的舆情事件数据,对后续舆情的发展态势作出判别;随后,引入分形插值理论实现舆情传播态势预测;最后,以某企业数字化转型舆情危机为例,验证其可行性。实验表明,该方法可对转型企业舆情事件进行实时监测和有效控制,辅助引导现代制造业转型过程中网络舆情事件的良性发展。

关键词: 现代制造业, 舆情监测, 多重分形, 分形插值, 态势预测