摘要
[研究目的]从视频弹幕的视角出发,挖掘网络舆情事件的话题漂移规律,提升网络舆情事件的视频情感检索精度。[研究方法]通过对视频弹幕进行主题与情感分析,提升网络舆情事件在线监测精准度,并在此基础上提出并构建弹幕迁移指数,建立一种基于弹幕迁移指数的情感监测方法,该方法首先基于BTM主题模型抽取视频弹幕的话题信息,并基于情感词典与颜文字词典计算不同时间窗口下的话题情感类别与情感强度,建立面向视频弹幕的网络舆情事件监测模型,再从话题内容的变化与视频兴趣热度两个角度构建话题迁移指数,并利用话题的情感强度变化,构建情感迁移指数。最终,基于话题迁移指数与情感迁移指数,得到加权后的弹幕迁移指数,实现网络舆情事件的在线监测。[研究结论]通过视频弹幕社区的真实数据,从逻辑层面验证了本模型的合理性,结果表明该方法能够较为准确地识别网络舆情事件迁移的关键时间窗口,为实现视频分享平台的情感可视化提供了切实可行的理论探索。
[Research purpose]From the perspective of bullet-screen video portrait,the topic sentiment and its evolution rules of network incidents are excavated,which improves the accuracy of video retrieval of network incidents.[Research method]The online image of network incidents is improved,through the topic and sentiment analysis of bullet screen videos.Afterwards,the bullet-screen drift index is proposed.To be specific,the topic information of bullet screen videos is extracted based on the BTM topic model.And the topic sentiment polarity and sentiment intensity under different time windows are calculated based on the sentiment dictionary.As a result,the network incidents portrait model oriented to the bullet screen video is established.Then,the topic drift index is constructed from two perspectives,namely the change of topic content and the video interest heat.Meanwhile,the sentiment drift index is constructed by using the change of topic sentiment polarity.Finally,the weighted bullet-screen drift index is obtained to realize online identification of the sentiment evolution of network incidents,on the basis of the topic drift index and sentiment drift index.[Research conclusion]The rationality of this model is verified through the real data of bullet video community users.The results show that this method can identify the critical time window of sentiment evolution of network incidents more accurately,which provides a practical theoretical exploration for realizing the sentiment visualization of video sharing platform.
作者
黄立赫
石映昕
Huang Lihe;Shi Yingxin(School of Marxism,Northwestern Polytechnical University,Xi'an 710129;School of Marxism,North China University of Water Resources and Hydro-power,Zhengzhou 450046)
出处
《情报杂志》
CSSCI
北大核心
2022年第2期146-154,共9页
Journal of Intelligence
基金
河南省高等学校重点科研项目资助计划“大数据时代河南高校图书馆数字资源综合服务绩效多维评估与优化”(编号:19A 870002)。
关键词
视频弹幕
网络舆情事件
弹幕迁移指数
颜文字词典
情感监测
bullet-screen video
network incidents
bullet-screen drift index
emoji dictionary
sentiment evolution
