| dc.description.abstract | Podcasts have gone mainstream and continue to attract ever larger audiences, particularly among younger demographics, while the internet-broadcast nature of the medium leaves listeners with little recourse in the face of potential misinformation. This work explores the feasibility and practicalities of transcribing and fact-checking podcasts using existing automated fact-checking technologies. A multi-purpose web application is constructed to store and display podcast transcriptions, collect human annotations from crowdsourcing platforms, and ultimately output a dataset useful for training and fine-tuning the next generation of fact-checking models to address the unique characteristics of podcast content. | |