流行度演化分析与预测综述

发布时间：2019-06-29 21:52

【摘要】：社交网络每天以爆发式的增长速率产生着大量信息,但是人们对海量信息的关注程度有限。人们关注哪些信息、对信息的关注程度如何随时间变化,即为信息的流行度演化问题。流行度演化反映了人们的关注点和信息的流动与传播。建模与预测网络信息的流行度演化有助于信息传播和人类行为的研究、辅助舆情监控、并带来极大的应用和商业价值。近几年,研究人员在该方面取得了丰硕的研究成果,但尚缺乏对这些成果进行梳理、总结的综述。该文系统地回顾网络信息流行度演化的主要工作,对分析与预测方法、模型、发展脉络进行梳理。首先从定性和定量方面阐述了流行度演化的特点;介绍如何量化影响流行度演化的众多因素,并对它们进行分类、总结;然后将已有的建模和预测方法归纳为3类:基于早期流行度、基于影响因素、基于级联传播,从原理、典型成果、特点比较、适用范围等方面对这3类方法进行评述;最后根据目前模型和方法的特点以及现实需求,指出了未来流行度演化的研究方向。
[Abstract]:Social networks produce a lot of information at an explosive growth rate every day, but people pay limited attention to the huge amount of information. What kind of information people pay attention to and how the degree of attention to information varies with time is the evolution of the popularity of information. The evolution of popularity reflects people's concerns and the flow and dissemination of information. Modeling and predicting the evolution of the popularity of network information is helpful to the study of information dissemination and human behavior, assists the monitoring of public opinion, and brings great application and commercial value. In recent years, researchers have made fruitful research results in this area, but there is still a lack of a summary of these results. This paper systematically reviews the main work of the evolution of network information popularity, and combs the analysis and prediction methods, models and development context. This paper first expounds the characteristics of epidemic evolution qualitatively and quantitatively, introduces how to quantify the many factors that affect the evolution of epidemic degree, classifies and summarizes them, and then classifies the existing modeling and prediction methods into three categories: based on early epidemic degree, based on influencing factors, based on cascade communication, from the aspects of principle, typical results, characteristic comparison, scope of application and so on, and then classifies the existing modeling and prediction methods into three categories: based on early epidemic degree, based on influencing factors, based on cascade communication, from the aspects of principle, typical results, characteristic comparison, scope of application and so on; Finally, according to the characteristics and practical needs of the current models and methods, the research direction of the evolution of popularity in the future is pointed out.
【作者单位】：北京科技大学计算机与通信工程学院;
【分类号】：TP393;;TP391

【相似文献】