网络事件扩散规律与早期预测研究

发布时间：2018-01-26 16:18

本文关键词： 网络事件网络舆情扩散规律扩散模型遗传算法　出处：《山东财经大学》2014年硕士论文　论文类型：学位论文

【摘要】：随着大数据、4G高速网络时代到来，实时交互、成本低廉的网络事件对社会生活、政策法规、人民心态影响逐渐加深。事件监控、舆论引导等问题引起广泛重视。本文针对网络事件发生的早期特征对其进行实时监测评价，对其未来发展进行预测，以期预警和引导舆论的良性发展。本文跟踪收集大量网络事件并记录早期特征和最终影响范围，全程评估及评价内容，归纳总结了网络事件扩散的时间规律、热度规律和生物时钟规律、不同网络事件信息筛选规律和同一网络事件信息筛选规律。本文通过实时网络事件归类、扩散性和倾向性检测，建立网络事件数据间的逻辑关系，，依据网络事件扩散规律建立了网络事件扩散模型。评价网络事件的影响力、实时监控网络事件的发展规模和方向，预测预警级别，确定舆论引导的方向和措施，在网络事件的发生初期进行引导，使其发展趋势及影响力在可控制范围内。本文基于遗传算法设计网络事件预测模型，在大数据环境下建立评价预测原型系统。对不同网络事件在同一时间发展情况进行预测，得到良好的实验结果，证明其使用价值与推广价值本文的主要创新点为： 1、发现黄金24小时为网络事件早期和快速上升期的分界点规律；提出在黄金24小时实时发现评价、实时预测引导的网络舆情监控方法，从事件发展为热点后的堵截转变为早期发现预测和良性引导。 2、基于网络事件内容监测和评论倾向性预测，建立自学习的网络事件扩散模型；利用遗传算法设计网络事件预测算法和原型系统，具有计算简单、处理时间快、自适应及规模、范围和态度控制的全面性等特点。
[Abstract]:With the arrival of big data 4G high-speed network era, real-time interaction, low-cost network events to social life, policies and regulations, people's mentality gradually deepened. Based on the early characteristics of network events, this paper carries out real-time monitoring and evaluation, and forecasts its future development in order to forewarn and guide the benign development of public opinion. This paper tracks and collects a large number of network events and records the early characteristics and final impact range, evaluates and evaluates the whole process, and summarizes the time law, heat law and biological clock law of network event diffusion. Different network event information screening rules and the same network event information screening rules. In this paper, the logical relationship between network event data is established by real-time network event classification, diffusion and tendency detection. According to the law of network event diffusion, the model of network event diffusion is established, the influence of network event is evaluated, the scale and direction of development of network event are monitored in real time, the warning level is predicted, and the direction and measures of public opinion guidance are determined. In the early stage of network events, the trend and influence of network events can be controlled. In this paper, a network event prediction model is designed based on genetic algorithm, and a prototype system of evaluation and prediction is established in big data environment. Different network events are predicted at the same time, and good experimental results are obtained. Prove its use value and popularize value The main innovations of this paper are: 1. It is found that gold is the dividing point between the early period of network event and the period of rapid rising in 24 hours. In this paper, a network public opinion monitoring method based on 24-hour real-time discovery evaluation and real-time prediction and guidance is proposed, which changes the interception from events to hot spots into early discovery, prediction and good guidance. 2. Based on the monitoring of network event content and the prediction of comment tendency, a self-learning model of network event diffusion is established. The genetic algorithm is used to design the network event prediction algorithm and the prototype system, which has the characteristics of simple calculation, fast processing time, adaptive and comprehensive control of scale, scope and attitude, etc.
【学位授予单位】：山东财经大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP18;TP393.06

【参考文献】