从我国网络舆情现状出发,分析网络舆情对社会影响的主要因素,选取4种网络舆情预警指标,获取观测数据,基于改进的D-S证据理论信息融合技术,建立网络舆情预警模型,并利用类概率函数实现对网络舆情话题演变成舆情事件的可能性的判断。同时把改进后的融合方法与经典的D-S组合规则做比较,结果表明,改进后的方法可以更好地实现证据的融合,并得到更为合理的判断结果。
According to the current situations of internet public opinion in China,we analyzed the main factors of the network public opinions which influence to the society,selected four indicators of early warning system,accessed to observational data,then established the warning model of internet public opinions based on information fusion technology of improved Dempster-Shafer evidence theory,and used the class-based probability function to realize the possibility of the network public opinion topic evolved into public opinion events.Meanwhile,the results show that the improved method can better achieve the combination of evidence,and more reasonable determined results are obtained by comparing the improved fusion method with the classic combination of D-S rules.