Description
Understanding the structure (e.g. source and types) and the trends of disinformation could provide decision support for related professionals (e.g. journalists, fact-checkers and policymakers) to prevent further dissemination of such information. To support such study, we developed an automatic classifier service that categorises COVID-19 misinformation into one of 10 categories, including government actions, impact on community, medical advice, conspiracy theories, vaccine, etc.
The dataset used comprises all English misinformation debunks published by IFCN until April 13, 2020, and manually labelled the debunks with predefined 10 COVID-19 classes based on Brennen et al., 2020 during the EUvsVirus hackathon.
- A service on GATE Cloud via the endpoint at https://cloud-api.gate.ac.uk/process-document/covid19-misinfo.
- A REST API at this endpoint: https://cloud-api.gate.ac.uk/process-document/covid19-misinfo.
Both the REST API and the online demo user interface return the same two different annotation types – MisinfoClass and Attention.