群体性突发事件的研究与预警实现

发布时间：2018-04-09 14:00

本文选题：群体性突发事件　切入点：预警　出处：《北京邮电大学》2012年硕士论文

【摘要】：随着我国改革开放的深化和市场经济的发展,很多新问题、新矛盾以不同的形式表现出来。特别是近年来我国群体性突发事件频频发生,造成的后果和影响也越来越严重,同时随着互联网的飞速发展,使得群体性事件的爆发更加突然,不断威胁到我国社会和谐,引起了社会各界的密切关注和深入思考。对于很多传统的群体性事件,可利用数据挖掘相关技术进行监控提取,但是近期出现的这样一类群体性事件,呈现出概率小、突发性、隐匿性等特点,其在产生之初,被掩盖在大量的平凡数据中,如用传统算法进行抽取,则会在数据处理初期被当成噪声数据滤除,从而没有引起足够的重视。如果忽视这类群体性突发事件,将会产生非常恶劣的后果。针对以上问题,本课题通过对这类群体性突发事件的研究,力图利用技术手段解决其发现预警不足的问题,通过整体建模和软件实现,期望能够很好的实现群体性突发事件的预警。同时,结合实际项目需求,将这个模型应用到基于某单位投诉文本的群体性突发事件分析预警系统中,用于预警某单位业务中出现的新问题和群体性事件。完成系统设计与实现的同时对下一步的深入研究也提出了一些建议。本文重点阐述了以下几个方面的工作： 1.通过分析得出这类群体性突发事件的定义、相关特性,并在此基础上提出了相应的数据模型,对群体性突发事件进行描述。 2.加强了对数据来源的分析,尤其是加入了网络数据作为数据来源的一部分,同时本文提出了一种基于滑动窗口模型的挖掘算法以去除噪声数据。 3.提出一种利用遗传算法优化的K-means聚类方法,以更好的实现聚类分析,提高了结果的准确性、可靠性。 4.将此分析预警模型应用到电信领域,结合与某单位合作的项目,对其相关投诉数据进行处理,实现对电信投诉的群体性突发事件预警。
[Abstract]:With the deepening of China's reform and opening up and the market economy, many new problems and new contradictions in different forms. Especially in recent years, the sudden mass incidents happen frequently in China, the consequences and impact is more and more serious, at the same time, with the rapid development of the Internet, the outbreak of mass incidents more suddenly and continue to threaten the harmony of our society, attracted close attention and deep thinking of the community. For many traditional group events, can use data mining technology to monitor the extraction, but the recent emergence of such group events, showing a small probability, sudden, characteristics of occult, the the beginning, hidden in the ordinary large quantities of data, such as the use of traditional algorithms for extraction, data processing will be in the early stage as noise data filtering, and if not attracted enough attention. The neglect of such mass emergencies will have very bad consequences.
To solve the above problems, through the research of this kind of group incidents, trying to use technical means to solve the warning problems, through the realization of integrated modeling and software, hoping to achieve mass emergency warning as well. At the same time, combined with the actual needs of the project, the model is applied to a group the text of the emergency Complaints Unit Based on the analysis of the early warning system for the event, new problems and group appear warning a unit in the business. At the same time to complete the system design and implementation of the next step of the research also puts forward some suggestions.
This article focuses on the following aspects of the work:
1., through the analysis, we get the definition and characteristics of such group emergencies, and on this basis, we propose corresponding data models to describe group emergencies.
2., we strengthened the analysis of data sources, especially joined the network data as a part of data sources. Meanwhile, we proposed a mining algorithm based on sliding window model to remove noise data.
3. a K-means clustering method optimized by genetic algorithm is proposed to improve the accuracy and reliability of the cluster analysis.
4., apply the analysis and early warning model to the field of telecommunications, and deal with the related complaints data combined with a cooperation project with a unit, so as to realize the early warning of mass incidents of Telecom complaints.

【学位授予单位】：北京邮电大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：TP311.13

【引证文献】