统计研究

• 论文 • 上一篇    下一篇

大数据背景下网络突发事件动态监测研究

唐晓彬等   

  • 出版日期:2017-02-15 发布日期:2017-03-01

Research on Monitoring Internet Burst Events Dynamically from the Big Data Perspective

Tang Xiaobin et al.   

  • Online:2017-02-15 Published:2017-03-01

摘要: 通过对互联网以及社交平台的数据监测社会突发异常事件是当前社交网络传播研究的热点问题,本文基于大数据背景下对网络突发事件动态监测算法进行了创新性研究。论文首先介绍了常用的几种的网络动态监测算法,并指出了其应用到社交网络中的局限性,提出了基于Kleinberg的改进算法,最后将改进的算法应用到“北京三里屯优衣库不雅视频”事件中,研究结果显示:改进后的算法通过给定合理的调节参数,能迅速准确的监测到网络异常状况的发生,从而不仅避免了传统主观上直接通过突发事件出现的频次来划定临界值的不科学性和直接使用微博量的排行榜方式所产生时间上的滞后性,也避免了直接通过划定一个固定的微博增加量来判断是否有异常事件发生的不合理性。本文在大数据背景下为网络突变事件的动态监测提供了新的研究方法,为政府今后加强网络安全建设、优化网络监管模式、净化互联网环境发挥一定的实践指导意义。

关键词: 大数据, 网络突发事件, 动态监测, Kleinberg算法

Abstract: Monitoring the social burst events by the data of social network site has been a hot subject in the research of social network communication currently. The paper puts forward an innovative algorithm to monitor the internet burst event from the big data perspective. Firstly, it introduces some usual detect methods and indicates the limitations of them when applied to social network. Then, it puts forward a modified algorithm basing on Kleinberg’s method. Finally, it applies the modified algorithm in the “Beijing SanlitunUniqlo indecent video” event. The results show that the modified algorithm can monitor the internet burst event quickly and accurately with a reasonable adjustable parameter. This method is more scientific than the traditional method which subjectively determines a critical value or fixed increment by the information frequency of the event. It also eliminates the lag of detecting burst event by the rankling list of the information frequency. Our method provides a new method for detecting and monitoring internet burst event dynamically from the big data perspective. Besides, it has great significance for the government to ensure the internet security, optimize the internet supervision mode and purify the internet circumstance.

Key words: Social Network Site, Burst Event, Dynamically Detecting, Kleinberg’s algorithm