On 10-12 January 2020, we held a fact checking hackathon for people to work collectively on the challenge of automated fact checking.
We kicked off with a few talks from people working in this challenging area. Mevan Babkar, head of automated fact checking at FullFact, gave an insightful talk about human-based fact-checking and their work during the recent general election. James Thorne gave a talk about his PhD work on verifying claims, which has included organising the Fact Extraction and VERification (FEVER) shared-task (http://fever.ai/). Jonty Page gave an overview of an open source fact-checking system the participants could work with.
Both Jonty and James were a massive help in the lead up to the event. Automated fact-checking is a complex task, but we managed to work out simpler work packages that participants from different background and with different skill sets could contribute to. A big thank is also needed for Dominic Friend and colleagues in the High Performance Computing, who helped set up accounts for all participants, pre-loaded with virtual environments with the necessary libraries and access to GPUs.
I was particularly keen to create a supportive and interdisciplinary context for collaboration. We managed to get participants from a really wide variety of contexts/disciplines (linguistics, psychology, sociology, education, criminology, mathematics, philosophy, natural language processing, computer science, and software engineering).
It was a great opportunity to come together with people from different backgrounds, people who are doing mathematics, engineering, computer science, linguistics, criminology … it was a really good environment to come together and get the different insights from the different perspectives that you wouldn’t necessarily think about
This is my first hackathon and I’ve really enjoyed its interdisciplinary nature, it’s really welcoming, it’s really engaging, it’s open to newcomers.
For a full write up of the event and to see videos from participants check out the CRASSH blog post.