基于本体的群体事件案例库数据挖掘应用研究

发布时间：2018-05-13 20:07

本文选题：数据挖掘 + 群体事件　；参考：《上海交通大学》2014年硕士论文

【摘要】：随着我国经济快速发展和社会的转型，近几年频繁出现的各类群体事件给人们的正常生产和生活带来了极大的困扰。而目前决策者们的处理效果却不甚理想，如何处理好群体事件和群体事件演化、发展的内在机理成为了众多学者研究的热点。与此同时，互联网的快速普及为人们提供了表达诉求的平台，网络舆情管理成为当下众多学者研究的热点。网络舆情的属性决定了网络舆情是把双刃剑，一方面能够帮助弱势群体解决诉求，另一方面也可能因为利用不当而造成更大的损失，研究网络舆情管理的重要性日益突出。由此，本文主要针对群体事件网络关注度与群体事件本身特征之间的关系展开研究和讨论，旨在为群体事件舆情预警和预判提供可行性建议。本文基于以上出发点，通过对群体事件典型案例的分析，，建立群体事件案例库表示框架，搜集自2010年以来的各类群体事件加入案例库，为群体事件案例分析和案例推理提供基础数据支持。其次，在学习和理解群体事件领域知识的基础上，本文建立了基于继承关系的群体事件本体，为后续数据挖掘提供了基础数据结构。再次，本文在传统ID3算法的基础上，融合继承关系本体，设计了基于本体的ID3改进算法，为群体事件网络关注度影响因素研究提供了工具；最后，运用基于本体的ID3改进算法，对群体事件网络关注度的影响因素进行了训练，得到了群体事件网络关注度的分类规则，通过规则的分析，了解群体事件网络关注度与群体事件本身特征之间的内在逻辑和联系，为群体事件舆情预警提供了可行性建议。
[Abstract]:With the rapid economic development and social transformation in China, the frequent occurrence of various group events in recent years has brought great problems to people's normal production and life. At present, the effect of the decision makers is not very good. How to deal with group events and the evolution of group events, the internal mechanism of development has become a hot spot of many scholars. At the same time, the rapid popularity of the Internet provides a platform for people to express their demands. The attribute of network public opinion determines that network public opinion is a double-edged sword. On the one hand, it can help the vulnerable groups to solve their demands, on the other hand, it may cause more losses because of improper use. The importance of studying the management of network public opinion is increasingly prominent. Therefore, this paper mainly focuses on the research and discussion of the relationship between the network attention of group events and the characteristics of group events, in order to provide feasible suggestions for public opinion early warning and prediction of group events. Based on the above starting point, through the analysis of the typical cases of group events, this paper establishes the case database representation framework of group events, and collects all kinds of group events since 2010 to join the case database. To provide basic data support for group event case analysis and case-based reasoning. Secondly, on the basis of learning and understanding the knowledge of group event domain, this paper establishes a group event ontology based on inheritance relation, which provides the basic data structure for subsequent data mining. Thirdly, based on the traditional ID3 algorithm, this paper designs an improved ID3 algorithm based on ontology, which provides a tool for the study of the influencing factors of network concern of group events. Finally, the improved ID3 algorithm based on ontology is used. The influencing factors of group event network attention are trained, and the classification rules of group event network attention degree are obtained. Through the analysis of the rules, Understanding the internal logic and relationship between the network attention and the characteristics of group events provides feasible suggestions for public opinion warning of group events.
【学位授予单位】：上海交通大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP311.13;D631.43

【参考文献】