基于UGC的网络视频舆情传播特征框架研究

Research on the Framework of Network Video Public Opinion Spread Feature Based on User Generated Content(UGC)

新媒体的发展开启了交互、多向的信息模式,极大地改变了人们的生活方式.网络舆论生态变得更为复杂,给网络舆论引导工作带来一系列现实的和潜在的问题.对此,提出了网络视频舆情传播特征(NVPOSF)框架,用于分析视频舆情特征.NVPOSF融合了模态内、模态间和双模态间的交互,构建了基于多头注意力的融合网络.NVPOSF通过为声学—视觉、声学—文本和视觉—文本特征分配合理注意力以获得重要特征.实验评估的结果表明,与现有的框架相比,NVPOSF具有更好的性能.为政府主动引领网络舆论,形成正面舆论强势,对社会重大事件的舆论引领起到重要作用.

The development of new media has opened up interactive and multi-directional information models,greatly changing people's lifestyles.The ecosystem of online public opinion has become more complex,bringing a series of practical and potential problems to the guidance of online public opinion.In response,the Network Video Public Opinion Spread Feature(NVPOSF)framework was proposed to analyze video public opinion features.NVPOSF integrates intra modal,inter modal,and bimodal interactions to construct a fusion network based on multi head attention.NVPOSF allocates reasonable attention to acoustic visual,acoustic text,and visual text features to obtain important features.The experimental evaluation results indicate that NVPOSF has better performance compared to existing frameworks.This actively leads online public opinion by the government,forming a strong positive public opinion and playing an important role in leading public opinion on major social events.