Reuters Executive Editor for Editorial Operations, Data and Innovation, Reg Chua, has told the press that the wire service has developed an algorithm to both identify and verify breaking news on Twitter. It is the latter that has proven problematic, which was perhaps most evident back on 23 April 2013 when the Twitter feed of Associated Press was hacked and a false report of a bomb scare at the White House caused a 1% fall in the S&P 500. The report was quickly discredited and markets bounced.
The software, which at present is only available internally, is designed as a support tool for the newsroom. Called "News Tracer", it has been in the making for over two years. It parses over half a billion tweets each day, filters them, sorts them into clusters, tags them by topic, then assigns them a 'credibility score'. This part does involve some manual intervention.
News Tracer systematically studies whether breaking news is being confirmed or denied by other Twitter users, but also references a database of accounts selected by Reuters' journalists. The extent to which other users are re-tweeted or followed by these handpicked accounts in part determines their credibility.
As yet, Reuters has not indicated if or when the tool will be made available to customers. At present, it is being used to help publish breaking news earlier than the competition and to allow journalists to focus on tasks "that machines don't do as well".