基于ISOA-SVR模型的短期网络舆情预测
Short-term network public opinion prediction based on improved seagull optimization algorithm and support vector regression model
杨赟 1张丽丽1
作者信息
- 1. 南京工业大学数理科学学院,江苏南京 211800
- 折叠
摘要
网络舆情传播具有时效性和小样本特征,提出一种改进海鸥算法优化支持向量回归的网络舆情预测模型ISOA-SVR.为提高SOA算法的性能,设计sigmoid函数非线性收敛因子实现种群迁徙与攻击阶段的平滑转换;引入精英个体多阶段动态扰动避免局部最优;设计正余弦优化指引种群位置二次更新,提高局部寻优能力.SVR学习效率高、逼近能力强,但对参数初值敏感、泛化能力仍有不足,利用ISOA算法对SVR优化调参,构建网络舆情预测模型ISOA-SVR.实验结果表明,ISOA-SVR数据拟合度更高,稳定性和收敛性表现更好.
Abstract
Aiming at the timeliness and small sample characteristics of network public opinion communication,a network public opinion prediction model based on improved SOA and optimized SVR(ISOA-SVR)was proposed.To improve the performance of SOA,a nonlinear convergence factor calculation method based on sigmoid function was designed to realize a smoother transfor-mation between seagull population migration and attack stage.The multi-stage dynamic disturbance mechanism of elite indivi-duals was used to avoid a local optimum.The sine cosine optimization was designed to guide the secondary update of population position and improve the local optimizing ability.The SVR show high learning efficiency and strong approximation ability,but there are still deficiencies in generalization ability.The improved SOA was used to optimize the support vector regression model and construct the network public opinion prediction model ISOA-SVR.The results show that ISOA-SVR not only has higher data fit,but performs better in stability and convergence speed.
关键词
网络舆情/支持向量回归/海鸥优化算法/sigmoid函数/多阶段动态扰动/正余弦优化/百度指数Key words
network public opinion/support vector machine/seagull optimization algorithm/sigmoid function/multi-stage dy-namic disturbance/sine cosine optimization/Baidu index引用本文复制引用
基金项目
江苏省高校哲学社会科学研究基金项目(2021SJA0219)
出版年
2024
NETL
NSTL
万方数据