摘要
Abstract
The purpose of this study is to deeply analyze the public's cognitive status of generative artificial intelligence(GAI),and provide data support and strategic suggestions for AI policy making and risk management.In this study,Weibo is selected as the data source,and a Zero-shot_M3E_BERTopic model is proposed to identify the public cognitive topics under GAI.Sentiment analysis is carried out in combination with ERNIE 3.0 model to quantify the topic's attention and satisfaction,and explore its dynamic change characteristics.Furthermore,based on the IPA model,a framework of"attention-satisfaction"analysis is constructed,and the public cognition is analyzed by the keyword co-occurrence network.The research finds that public cognition under GAI topic involves four dimensions and a total of 17 topics,in which Q&A quality,emotional insight,and educational assistance are the main vulnerable areas,while five topics such as job market,network security,and intellectual property rights are the secondary vulnerable areas.This paper gives the following suggestions from the perspectives of government,enterprises,and the public.The government should strengthen guidance and supervision,and build a solid support for development.Enterprises should accelerate technological innovation and build GAI with Chinese characteristics.The public needs to improve information literacy and use GAI tools rationally.关键词
生成式人工智能/IPA模型/关注度—满意度框架/BERTopic/ERNIE 3.0/主题聚类/情感分析Key words
Generative Artificial Intelligence/IPA Model/Attention-Satisfaction Framework/BERTopic/ERNIE 3.0/Topic Clustering/Sentiment Analysis分类
信息技术与安全科学
