一种基于网民行为的倾向性密度聚类方法

A tendency density clustering method based on netizen behavior

针对传统舆情检测方法存在检测指标单一、适用面窄、效率不高、预测滞后和结论不准确等缺陷,提出一种基于网民行为的倾向性密度聚类方法.采用网民行为与心智倾向的超大规模数据集,设计分布式架构的密度聚类算法.仿真结果表明,分布式聚类架构适用于超大规模的网民行为与心智倾向的预测需要,能够高效提取聚簇中心和实现仿真性链路预测目标,显著提高舆情新动向预测的实际功效.

Aiming at the defects of traditional public opinion detection methods such as single detection index,narrow application area,low efficiency,lagging prediction and inaccurate conclusions,a tendency density clustering method based on the netizen behavior was proposed.An ultra-large-scale data set of netizen behavior and mental tendencies was used to design a density clustering algorithm based on distributed architecture.The simulation results show that the distributed clustering architecture is suitable for the prediction of ultra-large-scale netizen behavior and mental tendencies,and can efficiently extract cluster centers and achieve simulation link prediction goals,significantly improving the actual effect of predicting new trends in public opinion.