Dataset for English Health-Related Advice Directed to the General Public on Twitter During the Early Spread of COVID-19 [Dataset]

Health-related advice directed to the public on twitterprovides insight into the use of social media duringa pandemic. This paper describes our data collection, sampling, and analysis of 44 million tweets in English in March 2020. We make reference to a parallel dataset and analysis of tweets in Arabic during thesame period. The contribution of this paper is a description of our dataset, our coding process to indicate tweets with health related advice, and our analysis and comparisons of the characteristics of the tweets with and without health-related advice. These contributions providethe basis for future research on semi-automated classifiers for health-related advice and efforts to reduce thespread of harmful health advice.面向公众发布的推特（Twitter）健康相关建议，可为疫情大流行期间的社交媒体使用场景提供研究视角。本研究详述了针对2020年3月发布的4400万条英文推文所开展的数据收集、抽样与分析工作。本研究同时提及同期阿拉伯语推文的平行数据集及对应分析内容。本研究的贡献包括：对本次数据集的详细说明、用于标注含健康相关建议推文的编码流程，以及针对含与不含健康相关建议的推文特征展开的分析与对比。上述研究贡献可为后续面向健康相关建议的半自动化分类器（semi-automated classifiers）研究，以及遏制有害健康建议传播的相关工作提供坚实的研究基础。