【作者】 张文斌；

【导师】 杨燕；

【作者基本信息】 西南交通大学 ， 计算机系统结构， 2011， 硕士

【摘要】 由于群体智能涌现过程和网络舆情发生过程具有的相似性,本论文提出将群体智能涌现应用于网络舆情预测系统的设计。网络舆情的及时发现有利于社会稳定。但是网络舆情的监测任务具有滞后性,使得在网络突发事件从爆发到被成功监测,然后被有效地控制和处理这期间,必然已经造成了不能挽回的损失和危害。因此本论文提出应该在网络突发事件爆发之前进行预测,即网络消息从发布到开始传播、或者在传播过程中,就应该被判断出是否会引发网络突发事件。群体智能涌现是在昆虫群体中发现的一种奇特现象,而群体智能涌现机理研究欲解决的问题是：“怎样的群体才能够表现出智能”。本论文结合相关参考文献以及作者对该问题的思考,总结了群体智能涌现的普遍特征,并对两大经典的群体智能算法进行了全新的诠释。智能个体的设计是智能群体设计的关键,而个体目标函数是智能个体设计的关键之一。本论文对从群体目标函数获取个体目标函数的方法进行了介绍,并将该方法运用到网络舆情系统中的虚拟群体的设计中,并对计算方法进行了改进——即在群体演化的过程完成个体目标函数的部分计算内容。最后,本论文设计并实现了基于群体智能涌现的网络舆情预测的原型系统,该系统根据网络博客作者的个性创建出仿真个体,构成仿真群体,并能够利用仿真群体的演化,对网络中可能引起真实网络用户群体重点关注的消息进行预测。更多还原

【Abstract】 Because of the similarity between the process of the emergence of swarm intelligence and the process of generating Webciety’s public sentiment, this thesis proposes to apply the emergence of swarm intelligence to the design of Webciety’s public sentiment forcast system.Discovering Webciety’s public sentiment in time is helpful for social stability. Monitering Webciety’s public sentiment constitutionally has a defect. When a monitering system successfully discovers a bombshell in Webciety, the harm may have already been generated, existing and transmited. So this thesis suggests forecasting the Webciety’s Bombshell before it erupted out. That means the forecasting system should estimate if a piece of news would become the trigger of a Bombshell, at the time the news just published on the Web, or during transmitting in the Web, in any case before the eruption of Bombshell.The emergence of swarm intelligence is an admirable phenomenon found from social insect. The research of the mechanism of the emergence of swarm intelligence is trying to solve such question:what swarm can performance advanced intelligence. The general characteristics of the emergence of the swarm intelligence are summed up, according to some references and the author’s ideas.The design of the intelligent individual is the key point of designing a intelligent swarm. And the individual’s private utility is one of the key of designing an intelligent individual. The method of obtaining individual’s private utility from the swarm’s global utility is introduced. Then the method is applied to design virtual swarm of the Webciety’s public sentiment forcast system. The computing method is improved as doing a part of computation of private utility while the swarm evoluting.Finally, a prototype system for forcasting Webciety public sentiment is designed and implemented. The prototype system is designed based on the theory of the emergence of swarm intelligence. The prototype system creates emulational agents according to Blog authors’personality. After the evolution of the emulational swarm consisted of the emulational agents, the messages that most deserve real Web user group’s attention were forcasted.更多还原