采用实时线性模型的微博话题预警分析

发布时间：2018-03-08 13:06

  本文选题：微博　切入点：实时　出处：《图书情报工作》2017年15期 　论文类型：期刊论文

【摘要】：[目的/意义]微博在当前信息传播中起着重要作用,为有效预测微博热点及舆情导控,建立实时线性预警模型。[方法/过程]将采集的指标进行缺失值和异常值的处理后,对微博话题热度与大V影响力因子进行因子分析与逐步回归的比较,筛选出公共影响因子;再对其加权,探索不同权重调节因子下的最佳定量公式;用此公式每次输入当前时刻起前3小时的数据,预测当前时刻起后30分钟的加权值对应的话题词,每隔10分钟重新更新一遍参数。[结果/结论]实验证明该预测模型能大大降低数据采集解析和预测时间,保持较好的准确率,并可通过选择合适的阈值,进一步提升精确度。
[Abstract]:[objective / significance] Weibo plays an important role in the current information dissemination. In order to effectively predict Weibo hot spot and public opinion guidance, a real-time linear early warning model is established. [method / process] after processing the missing and abnormal values of the collected indexes, The factor analysis and stepwise regression of Weibo's topic heat intensity and large V influence factor are compared, and the public influence factors are screened out, and the optimal quantitative formula under different weight regulating factors is explored. The formula is used to input the data of the first 3 hours from the current moment each time, and to predict the topic words corresponding to the weighted value of 30 minutes from the current moment. The experimental results show that the prediction model can greatly reduce the time of data acquisition, analysis and prediction, and maintain a good accuracy, and the accuracy can be further improved by selecting the appropriate threshold.
【作者单位】： 韶关学院信息科学与工程学院;广西师范大学数学与统计学院;
【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)研究成果之一
【分类号】：G206;TP393.092
，

本文编号：1584016