论文部分内容阅读
【目的】针对现有预警体系多以企业自身和监管部门为主体、忽视网络舆情,导致预警力度不强、缺乏透明度及敏感性、使突发性安全问题时有发生且无法得到及时处理的现状,提出一种新的舆情预警模型。【方法】通过元搜索技术挖掘舆情信息,增加基准偏移值优化情感特征项倾向性权重,添加修正因子以改进潜在语义分析和支持向量机(LSA+SVM)算法,构建舆情分类预警模型。【结果】以多组突发性安全事件为例,应用Matlab进行仿真实验。结果证明该舆情预警模型切实可行,反应迅速,在语义维度为10时准确率可达85.75%。【局限】此方法对于能引起关注和讨论的安全事件更加有效。【结论】改进算法适用于舆情预警,可为企业和监管部门根据分类结果及时采取有效的预警措施提供合理化建议。
【Objective】 In view of the fact that the existing early-warning system mainly focuses on the enterprises themselves and the regulatory authorities, ignoring the public opinion on the Internet, resulting in weak early warning, lack of transparency and sensitivity, and making sudden security problems happen at any time and can not be dealt with promptly , Proposed a new early warning model of public opinion. 【Method】 The meta-search technique was used to mine the public opinion information, the baseline offset value was added to optimize the preference weight of emotion feature item, and the correction factor was added to improve the latent semantic analysis and support vector machine (LSA + SVM) algorithm to construct the early-warning model of public opinion classification. [Results] Taking multiple groups of emergent safety incidents as an example, the simulation experiments were carried out using Matlab. The results show that the public opinion early warning model is feasible and responsive, with an accuracy rate of 85.75% when the semantic dimension is 10. [Limitations] This method is more effective for security incidents that can cause concern and discussion. 【Conclusion】 The improved algorithm is suitable for early warning of public opinion, which can provide reasonable suggestions for enterprises and regulatory authorities to take timely and effective precautionary measures according to the classification results.