摘要
网络平台中舆论主题演化、起伏的周期与节律是否以及如何影响到主题的热度,是具有困惑度和缺乏研究的问题,探索该问题对于研究主题周期功能、主题演化机制等具有重要意义。本文基于今日头条平台帖子样本,探索主题的周期长度对主题热度的作用及可预测模型。采取功率谱分析计算主题周期长度,使用逻辑回归与决策树检验主题周期长度对主题热度的预测效果。结果表明:1)多数主题具有1周、1/3周、1/2周这三种周期,可称为平台三大“主频”,“主频”对主题热度起到抑制作用;2)对主题热度起到抑制作用的还有1/2周左右的“波长”集群、1周左右的“波长”集群;3)[2.6天,3天]区段(α区段)、[5.2天,6天]区段(β区段)内的周期长度,对主题热度起正向作用,它们主要是处于6天长度内的“快频率”和短波长,且在三大“主频”的抑制区波长以外;4)α区段约处于1/2周和1/3周的这2个主频中间,β区段约处于1/2周和1周这2个主频中间,且α区段与β区段之间有近似的2倍关系。上述周期长度的特征,可对主题热度起到一定的预测作用。
It is a problem with confusion and lack of research whether and how the evolution,fluctuation cycle and rhythm of public opinion in the network platform affect the popularity of the topics.Exploring this problem is of great significance for studying the function of topics cycle and the mechanism of topics evolution.Based on post samples of Toutiao platform,exploring the effect of the period length of the topic on the popularity of the topic and its predictable model in this paper.Power spectrum analysis was used to calculate topic cycle length,and logistic regression and decision tree were used to test the prediction effect of topic cycle length on topic popularity.The results show that:1)Most themes have a cycle of 1 week,½week,1/3week,which are called the prominent"main frequency"of Toutiao platform,and these three cycle lengths have a negative effect on the popularity of the theme;2)The"wavelength"cluster of about⅓week and 1 week also have a negative effect on the popularity of the theme;3)The period length in the[2.6 day,3 day]segment(αsegment)and[5.2 day,6 day]segment(βsegment)have a positive effect on the subject heat,which are mainly in the"fast frequency"and short wavelength within the 6-day length,and outside the wavelength of the above three categories of suppressed regions;4)Theαsegment has a positive influence on the frequency of1/3and1/2weeks,and theβsegment is in the middle of the 2 frequencies of1/2weeks and 1 weeks,and there is an approximate 2x relationship between theαsegment and theβsegment.Through the characteristics of the above cycle length,the theme popularity can be predicted to a certain extent.
作者
徐翔
杨心茹
XU Xiang;YANG Xinru(College of Arts&Media,Tongji University,Shanghai 201804,China)
出处
《中国传媒大学学报(自然科学版)》
2024年第3期1-9,共9页
Journal of Communication University of China:Science and Technology
基金
国家自然科学基金项目(71804126)
上海市“科技创新行动计划”软科学研究项目(23692110600)。
关键词
网络平台
周期长度
传播热度
功率谱
network platform
cycle length
propagation popularity
power spectrum