论文部分内容阅读
[目的/意义]突发公共卫生事件的迅速蔓延性与高度危害性使得相关管理部门不得不在短时间内做出有效的应对措施。对突发公共卫生事件的微博影响力进行预测有助于及时发现即将出现的问题,提高决策的预见性。[方法/过程]从微博条目的转发、评论和收藏次数来衡量其影响力,利用基于BM25的潜在狄利克雷分配模型和随机森林方法,选取埃博拉爆发相关微博的发布者、时间和内容特征,构建了突发公共卫生事件的微博影响力模型。[结果/结论]该模型能较好地预测微博的影响力,准确率达到88.8%。微博条目的发布者类型、主题等特征对微博条目的影响力产生主要作用,各项特征具有不同的影响力倾向。
[Purpose / Significance] The rapid spread of public health emergencies and the high degree of danger make the relevant management departments have to make effective response measures in a short period of time. Forecasting the impact of microblogging on public health emergencies helps to detect impending problems in time and improve the predictability of the decisions. [Method / Process] Weighing the impact of Weibo entries by the number of forwarding, commenting and collecting times, we use the potential Dicklecley distribution model based on BM25 and random forest method to select the publisher of the microblogging related to Ebola outbreak, And content features, we constructed a model of microblog influence on public health emergencies. [Results / Conclusion] The model can predict the impact of Weibo better, with an accuracy rate of 88.8%. The characteristics of the microblog entries, such as the type and subject of the publisher, have a major impact on the influence of the Weibo entries, and each feature has a different tendency of influence.