基于可解释性GAN与ChatGPT协同的网络健康谣言识别研究

打开文本图片集

关键词：可解释性GAN；ChatGPT；健康谣言识别；动态反馈优化机制；跨模型特征融合 DOI：10.3969/j.issn.1008-0821.2026.04.016 [中图分类号]R-05；TP391.1；G206 [文献标识码]A[文章编号]1008-0821（2026）04-0183-15

Abstract：[Purpose/Significance]The proliferation of online health rumors misleadspublic perception，intensifies social panic，andeven endangers publiclifeandhealth，posinga keychaengetopublic health governance in thedigital age.The paper proposes areliable andadaptable technicalapproachforthe intelligent governanceofhealth rumors，which isofgreat practical significance for buildingahealthyinformationecosystemandenhancingpublicscientific literacy. [Method/Process]The paper proposed the EGAN-GPT model based on the collaboration of explainable GAN and ChatGPT，which improved the performance of online health rumor identification through cross-model feature fusionand dynamic feedbackoptimizationmechanisms.Experiments utilizedonline health informationanddebunked health rumor data，comparing theidentificationperformance oftheEGAN-GPTmodelwitheightbaseline models onthetestset.Result/ Conclusion] The experimental results show that the EGAN-GPT model achieves an identification accuracy of 91.6% and an F1 score of 91. 5% foronline health rumors，with an average improvement of6 percentage points over the baseline models. It also demon-strates significant advantages in explainability，robustness，and cross-scenario adaptability.

Keywords：interpretableGAN； ChatGPT；health rumor identification；dynamic feedback optimization mechanism; ：ross-model feature fusion

全球健康谣言的爆发式增长，导致公众对卫生系统出现信任危机[1]，并对人们的日常生活产生了巨大的影响。（剩余21634字）