Scientists publish literally thousands of journal articles each month in all sorts of different fields. Linking all that information together is a mammoth task, but that's where CRAB comes in. It reads papers, finds links and highlights cancer causes.
A system developed by Cambridge University and the Karolinska Institute in Sweden; CRAB's special because it reads the scientific papers like a human would, but much, much faster. Using natural language recognition it's able to pull together information that would otherwise remain unlinked, due to the sheer scale of the amount of information spat out by science. CRAB's primary task will be to text mine for possible carcinogens, or cancer causing chemicals, scanning thousands of papers on each chemical and pointing out any potential dangers.
The aim is to make CRAB available to researchers via an online toolset for risk assessment. Speeding up risk assessments is great, but If they manage to expand its remit towards more general search terms, it could have enormous potential to speed up science. Hopefully it could render the manual scouring of journals moot, and find links scientists wouldn't be able to on their own. I would have killed for something automated like that during my doctorate, and if it proves successful, CRAB could reduce the time taken to thoroughly search the scientific record for a specific topic by months, if not years. [CRAB, The Telegraph]