摘要
为了提高微博舆情的预测精度,针对不同单一核函数的局限,用线性拟合确定两种核函数的权重提出改进的支持向量机模型。首先利用马尔科夫模型矩阵的稀疏程度提取影响因子指标,得到微博传播的增减趋势;然后用改进的支持向量机对实时数据按照4∶1的比例划分测试集和训练集,进行实时预测与警示。实验结果表明:应用马尔科夫模型进行微博舆情的主成分提取效果较佳,改进的支持向量机构造了新的组合核函数,比传统的预判效果更佳。
In order to improve the prediction accuracy of Microblog public opinion and make up for performance deficiency of single kernel function,the weight coefficients of two kernel functions have been calculated by linear fitting. The Markov matrix was used to determine the weights of the impact factors and the trend of Microblog public opinion. Improved support vector machine was used to divide real time data into training set and test set according to the proportion of 4∶ 1.Experiment showed that the features which affected micro blogging publica opinion,had been mined better by using Markov model; Optimized SVM model constructed a new combined kernel function,and the forecasting results were better.
作者
饶浩
文海宁
林育曼
陈晓锋
Rao Hao Wen Haining Lin Ymnan Chen Xiaofeng(Department of Information Management, Shaoguan University, Shaoguan 512005, China School of Mathematics and Statistics, Guangxi Normal University, Guilin 541004, China Department of Educational Technology, Shaoguan University, Shaoguan 512005, China)
出处
《现代情报》
CSSCI
北大核心
2017年第3期46-51,共6页
Journal of Modern Information
基金
教育部人文社会科学研究项目"社交媒体潜在舆情发现及导控机制研究"(项目编号:13YJCZH144)
广东省哲学社会科学规划项目"基于社交媒体的移动学习研究与实践"(项目编号:GD13CJY07)
广东省攀登计划项目"大学生微博热点话题趋势预测系统"(项目编号:pdjh2015a0471)
关键词
马尔科夫模型
组合支持向量机
微博
舆情
热点话题
预测
Markov model
combination support vector machine
microblog
public opinion
hot topic
prediction
