振荡的琴弦:周期特性对网络主题触发能力的影响及预测  

Vibrating Strings:The Influence and Prediction of Cyclical Characteristics on the Triggering Ability of Topics

作　　者：徐翔 宋雨萱 Xu Xiang;Song Yuxuan(College of Arts and Media,Tongji University)

机构地区：[1]同济大学艺术与传媒学院,上海201804

出　　处：《新闻界》2025年第3期50-68,共19页

基　　金：国家自然科学基金项目“社交网络互动中用户‘信息窄化’机理分析:基于微博的数据挖掘”(71804126);上海市软科学研究项目“从各有所好到共同牢笼:推荐算法平台公共信息茧房形成机理与调控策略”(23692110600)。

摘　　要：网络主题的周期性振荡特点及其所反映的时间模式,是否以及如何作用于主题之间的触发,是尚未明确的理论问题。主题的周期性振荡特点不只是反映在“肉眼可见”的峰-谷间的起伏波形,也反映在“肉眼不可见”地对起伏波形拆解后的若干种振荡频率、振荡波长及其振荡强度的组合,它们反映主题振荡、循环的运动形式和时间模式。各个主题犹如一根根具有自身各种振荡波长及其对应的振荡强度、形成特殊振荡模式的“琴弦”,在舆论的触发、共振和联动中起着独特作用。本研究借助功率谱分析与红噪声检验的频域分析方法,提取主题的显著振荡频率及其对应的周期长度、在特定周期长度上的振荡强度,通过检验四个假设与构造XGBoost回归预测模型,检验主题周期性振荡特点对主题触发能力的影响。结果发现,网络主题周期性振荡频率的集中度、总振荡程度、平均振荡程度越高,主题触发能力越强;主题周期性振荡程度在各显著频率分布越不均衡,主题触发能力越强。这四个周期特征能有效预测主题触发能力。此外,主题内容相似度对预测主题触发能力并无足够贡献。较之信息“狂轰滥炸”的简单大规模复制,主题信息以何种节律、模式来振荡具有重要性和技巧性,关系到信息有效地传播、激发和调控,可能而且需要把议程设置理论中强调的“议题内容设置”扩展到“议题时间形式设置”。网络主题的周期性振荡特点为研究主题间联动触发、协同演化提供了新视角,有助于掌握网络中的节律、进行有效的舆论调节。The periodic oscillation characteristics and its time pattern of online topics and their impact on the triggering of inter-topic interactions remain an unclear theoretical issue.The periodic oscillation characteristics of the topic are not only reflected in the visible peaks and valleys of the undulating waveform,but also in the combination of several oscillation frequencies,oscillation wavelengths,and oscillation intensities that are not visible to the naked eye after disassembling the undulating waveform.These characteristics reflect the motion forms and time patterns of topic oscillations and cycles.Each theme is like a string with its own oscillation wavelengths and corresponding oscillation intensities,forming special oscillation modes,playing a unique role in triggering,resonating,and linking in public opinion.This study employs power spectrum analysis and red noise test frequency domain analysis methods to extract significant oscillation frequencies and corresponding cycle lengths of topics,by testing four hypotheses and constructing a XGBoost Regression prediction model,we examine the impact of periodic oscillation characteristics on the triggering ability of topics.The results indicate that the higher the concentration,total oscillation degree,and average oscillation degree of periodic oscillation frequencies,the stronger the triggering ability of the topic.Additionally,the more uneven the distribution of periodic oscillation degrees across significant frequencies,the stronger the triggering ability of the topic.These four periodic characteristics effectively predict the triggering ability of topics.Moreover,the similarity in topic content does not contribute enough to predicting the triggering ability of topics.Compared to the simple large-scale replication of information bombardment,the rhythm and pattern in which thematic information oscillates is of great importance and skill,which is related to the effective dissemination,stimulation,and regulation of information.It is possible and necessary to exte

关 键 词：周期特征 主题触发 功率谱 红噪声 XGBoost回归 格兰杰因果分析 

分 类 号：G206[文化科学—传播学]