中文摘要：

人们在网络上发表的对现实生活中事件的观点，并彼此影响，特别是对突发事件的表达将引领社会思潮。因此对网络舆情的跟踪、监测以及引导具有重要的现实意义。越来越受到业界与政府部门的关注。而网络舆情发展演化的预测尤为重要。目前在突发事件预测前缺乏有效手段来判断拟采用的舆情演化模型的有效性，且其参数主要依靠经验设定。因此，借鉴石油开发中油藏数值模拟技术，将历史拟合引入到舆情分析中。通过对所监控社区的历史事件进行拟合，来确定所使用的网络舆情演化模型的有效性，并获取符合真实情况的模型参数。同时，提出了基于粒子群的舆情历史拟合算法。实验表明，该算法能够获得较好的结果。

英文摘要：

In recent years, people and government have paid more and more attention on public opinion research. Internet-users could freely comment on or express their opinion about the real-life events online, as influences each other and reflects the social thought. There-fore, the discovery, monitoring and guidance of public opinion have practical significance to the social life. In this case, it is particularly important to grasp the evolution trend of public opinion. However, due to the fact that parameters of the current public opinion evolution model mainly depend on empirical setting, it is a meaningful thing to find a method which gains parameters that can reflect the reality of monitored area. The paper introduces the history matching into the evolution of public opinion analysis, then, obtains the corresponding parameters by it, and presents a history matching algorithm based on the particle swarm algorithm. Experiment result shows that the algo-rithm can effectively get the model parameters of public opinion, and achieve good evolution results.