ChatGPT在健康谣言鉴别中的实证探讨与应用展望
An Empirical Discussion and Application Prospect of ChatGPT in the Identification of Health Rumors
潘雪峰 1王超 1卢智增2
作者信息
- 1. 辽宁工业大学图书馆 辽宁锦州 121000
- 2. 桂林理工大学公共管理与传媒学院 广西桂林 541000
- 折叠
摘要
[目的/意义]探讨ChatGPT在健康谣言鉴别中的应用效果,为预训练大模型在健康信息领域的广泛应用提供参考.[方法/过程]研究者从权威辟谣网站上选择健康信息,使用ChatGPT鉴别网络健康信息的"真""伪",并与机构专家鉴定结果进行比较.[结果/结论]ChatGPT大模型鉴别准确率为 93.1%,生成的解释性文本内容详细,语言流畅,对健康信息的理解在很大程度上接近医学专家.ChatGPT大模型具有良好的鉴别效果,但需要人工干预以确保结果的准确性和可靠性.未来在人工智能大模型研究领域,研究者应重视垂直领域高质量语料库的构建以及针对不同领域进行微调和优化.
Abstract
[Purpose/significance]This paper discusses the application effect of ChatGPT in the identification of health ru-mors,and provides reference for the wide application of pre-trained large model in the field of health information.[Method/process]Researchers select health information from authoritative rumor-refuting websites,use ChatGPT to identify the"true"and"false"of on-line health information,and compare it with the identification results of institutional experts.[Result/conclusion]The identification ac-curacy of ChatGPT model is 93.1%.The generated explanatory text is detailed,the language is fluent,and the understanding of health in-formation is close to that of medical experts to a large extent.ChatGPT large model has a good discrimination effect,but requires manual intervention to ensure the accuracy and reliability of the results.In the field of artificial intelligence large model research in the fu-ture,researchers should pay attention to the construction of high-quality corpora in vertical fields and fine-tune and optimize them for different fields.
关键词
健康信息/鉴别方法/人工智能Key words
health information/identification method/artificial intelligence引用本文复制引用
基金项目
国家社会科学基金一般项目(20BKS009)
出版年
2024
NETL
NSTL
维普
万方数据