网络舆情事件的主动感知实践
2015-12-14黄炜姚嘉威
黄炜++姚嘉威



〔摘要〕随着网络时代的到来,网络数据呈指数爆炸式增长,主题的模糊性越来越明显。同时多元非结构性的数据使得传统的聚类算法在网络舆情事件的发现越来越困难,不能满足高效,精准,及时、有效的感知需求。本文引入LDA聚类算法,基于主题生成模型,挖掘数据背后的语义关联,设计并且实现舆情事件的热点主动感知系统。通过数据实验表明,该系统能够快速、高效地发现事件主题,克服偏移词的干扰,从而实现网络舆情事件热点的主动感知。
〔关键词〕网络舆情;热点事件;LDA;聚类
DOI:10.3969/j.issn.1008-0821.2015.10.002
〔中图分类号〕TP391〔文献标识码〕A〔文章编号〕1008-0821(2015)10-0007-05
Research on Detection of Network Public Opinion EventHuang Wei1,2Yao Jiawei1
(1.School of Economy and Management,Hubei University of Technology,Wuhan 430068,China;
2.School of Management,Wuhan University of Technology,Wuhan 430070,China)
〔Abstract〕With the era of cloud computing and data arrival,the amount of data the exponential explosion,ambiguity and complexity increase and the theme of the more obvious,and massive multiple non-structured data,the traditional clustering algorithm is found and perceived significantly more and more limitations in the event of network public opinion,can not meet the high efficiency,accurate,timely,effective demand.This paper introduced the modern LDA clustering algorithm,which was based on the theme of generation model,capable of semantic association mining behind the data,through the continuous evolution of reasoning,in order to explore the data hidden value,design and implementation of public opinion events hot perception system.Through a large number of experimental data obtained,the system could efficiently and quickly found the data subject,accurately grasp the core essentials,and ignore the interference of individual words,so as to determine the perception of Internet public opinion hotspot.
〔Key words〕network public opinion;hot topic event;LDA;clustering
网络信息的爆发式增长,传统的分析方法已经不能适用这样的环境。很多垃圾信息充斥着互联网,导致越来越多的信息资源并没有被人们所利用。与此同时,泛在网络和自媒体的快速发展正改变着传统信息传播的媒介和方式,凭借其开放性、实时性和自由性,迅速占领了网络应用市场,例如微博和微信。人们利用这些工具进行随时随地的信息发布和传播,从而使社会的各种矛盾通过网络不断展现和放大,产生了较大的负面影响。如何及时发现和感知人民群众所关心和关注的热点,促进网络文明社会的和谐发展,在新形势下显得格外重要。
基于主题发现的LDA[1]文本聚类在此需求下应运而生,旨在辅助用户快速有效地找到所需资源,提供更精准的主题信息服务。本文引入LDA聚类算法进行海量数据处理,分类汇总,提炼数据背后的关联主题,从而提高信息检索和主题发现的精准率,为准确把握网络舆情事件的热点主题服务。……
