摘要
ChatGPT是基于预训练的生成式对话模型,通过深度学习技术实现对人类语言的理解和生成能力,因其通过大规模的语料库训练,能够自动生成连贯、富有逻辑的对话回复而在各领域都产生了巨大的轰动,尤其是大众重点关注的教育领域。通过网络爬虫技术挖掘视频网站的相关弹幕信息,使用WordCloud工具绘制词云图,并使用中文文本情感分析工具SnowNLP进行情感得分统计并将结果可视化,接着使用潜在狄利克雷分配(Latent Dirichlet Allocation,LDA)模型对文本主题进行挖掘并分析,研究发现大众对于ChatGPT在教育领域的社会影响、家庭教育、人类思维和学校教育等方面表达出了较为积极的态度,对于ChatGPT实名制问题、企业运营、人类就业和交互体验等方面表达了担忧。
ChatGPT is a pre-trained generative conversational model that utilizes deep learning techniques to understand and generate human language.Due to its ability to generate coherent and logically consistent dialogue responses through training on large-scale corpora,it has gained tremendous attention across various fields,particularly in the domain of education.In this study,we employed web crawling techniques to extract relevant bullet screen information from video sharing platforms,used the WordCloud tool to visualize word frequency distributions and conducted sentiment analysis using the Chinese text sentiment analysis tool SnowNLP.The sentiment scores were statistically analyzed and visualized.Additionally,we applied the Latent Dirichlet Allocation(LDA)model to uncover and analyze the underlying themes in the text.The research findings indicate that the general public expresses a positive attitude towards the social impact of ChatGPT in the field of education,as well as its implications for family education,human thinking,and school education.However,concerns were also expressed about issues such as real-name system implementation,corporate operations,human employment,and user interaction experience related to ChatGPT.
作者
刘兴红
张苏薇
王瑰霞
余亚烽
李冰楠
LIU Xinghong;ZHANG Suwei;WANG Guixia;YU Yafeng;LI Bingnan(College of Computer and Information Engineering,Hubei Normal University,Huangshi 435002,China)
出处
《湖北师范大学学报（自然科学版）》
2024年第4期52-58,共7页
Journal of Hubei Normal University：Natural Science
基金
2023年湖北师范大学研究生创新科研项目(2023Y042)
2022年湖北省教育厅哲学社会科学研究项目(22Y105)。
