基于知识发现的突发事件舆情风险演化模型研究

The Research on the Model of Risk in Emergency Online Public Opinion Bases on Knowledge Discovery

【目的/意义】为增强舆情风险应对能力、提高舆情识别和演化的准确性,本研究提出并构建基于知识发现的突发事件舆情风险演化模型。【方法/过程】结合舆情风险生命周期与社会燃烧理论,从燃烧物质和助燃剂及导火线角度,将舆情风险演化过程划分为萌发阶段、扩散阶段、爆发阶段和消退阶段,并分析了各阶段中关键属性和属性间存在的紧密关联性。在此基础上,利用知识图谱表征现实世界中的各种定义与常识,构建关联规则结构语义知识库,并引入卷积神经网络,将其当作分类器以获取舆情识别模型。利用卷积神经网络中的知识发现法实现舆情情感识别,将输出结果与舆情风险演化流程中的各阶段相对应,更显著地表现突发事件舆情风险的演化过程。【结果/结论】实验结果显示该模型性能较好,能够能很好的拟合突发事件舆情整体的演化规律,具有较强的实用性。【创新/局限】同时,本次研究未将网民异质性考虑到舆情风险演化模型设计中,导致模型精度存在一定的局限,但这也为日后更进一步的研究提供了新的思路。

【Purpose/significance】In order to enhance the ability to deal with public opinion risk and improve the accuracy of public opinion recognition and evolution, this study proposes and constructs an evolution model of public opinion risk in emergencies based on knowledge discovery.【Method/process】Based on the life cycle of public opinion risk and the theory of social combustion, the evolution process of public opinion risk is divided into germination stage, diffusion stage, explosion stage and extinction stage from the perspective of combustion materials, accelerators and fuses. On this basis, the knowledge map is used to represent various definitions and common sense in the real world, and the semantic knowledge base of association rule structure is constructed, and the convolutional neural network is introduced as a classifier to obtain the public opinion recognition model. The knowledge discovery method in the convolutional neural network is used to realize the emotion recognition of public opinions, and the output results are corresponding to each stage in the evolution process of public opinion risk, so as to show the evolution process of public opinion risk in emergencies more significantly.【Result/conclusion】Experimental results show that the model has good performance, can well fit the overall evolution of public opinion in emergencies, and has strong practicability.【Innovation/limitation】This study did not consider the heterogeneity of netizens in the design of the public opinion risk evolution model, resulting in certain limitations in model accuracy, but this also provides new ideas for further research in the future.