Towards reliable automated fact-checking in Public Health

过去几年，随着假新闻在媒体生态系统中的扩散，人们对事实核查的兴趣越来越大。事实上，研究人员正在探索如何利用机器学习和自然语言处理的最新进展来自动化事实核查，以打击错误信息的传播。到目前为止，这一领域的大部分工作都集中在验证政治主张上，而对公共卫生几乎没有兴趣，尽管相信这一领域的错误信息可能会造成危害。本论文的目的是发展公共卫生声明的自动事实核查研究。我们将首先创建第一个数据集，用于针对此类声明的大型知识库进行事实核查。这将通过为社区提供探索该主题所需的数据来促进该领域的研究。从这个数据集中，我们希望探索许多令人兴奋的研究途径。我们提出了多种建模思想，最终建立了一个端到端的公共健康声明事实核查系统。我们希望，这项工作将标志着在打击公共卫生错误信息的令人兴奋的努力中取得重大进展。

Interest in fact-checking has grown over the past few years with the proliferation of fake news in themedia ecosystem. Indeed, researchers are exploring how fact-checking can be automated using thelatest advances in Machine Learning and Natural Language Processing to combat misinformationspread. So far, most of the work in this field has focused on verifying claims in politics with littleinterest in public health despite the harm that could result from believing incorrect information in thisdomain. The purpose of this thesis is to develop the research in automated fact-checking for publichealth claims. We would start by creating the first dataset for fact-checking against a large knowledgesource for such claims. This would facilitate the research in the field by providing the communitywith the required data to explore the topic. From this dataset, there are many exciting avenues forresearch that we wish to explore. We propose multiple modeling ideas that culminate in buildingan end-to-end fact-checking system for public health claims. We hope that this work will mark asignificant development in an exciting effort to combat misinformation in public health.