Extraction of Teletext Subtitles from Broadcast Television for Archival and Analysis

Teletext subtitles are broadcast in some regions of the world, providing a potential resource of textual information corresponding to television current affairs, talk and magazine shows. Such programming offers an insight into sentiment, and will be useful in studies of national and global events. This paper discusses how textual information may be derived from digital television broadcasts featuring teletext subtitles, and shows some basic examples of their analysis for keywords related to the COVID-19 pandemic. The method has further possibilities, such as for the development of a searchable national archive of broadcast materials for education and research, and development of datasets for topics such as machine learning in sign language and sentiment analysis.