Reuters launched an in-house tool that brings automation into its journalism, lending a helping hand to its reporters to find more stories in data, the newswire announced today (12 March).
The tool, called Lynx Insight, is the latest launch as part of the organisation's drive to become a "cybernetic newsroom, marrying the best of machine capability and human judgement to drive better journalism," Reg Chua, executive editor, editorial operations, data and innovation at Reuters, wrote in a blog post announcing the introduction of the tool.
Lynx Insight is rolled out with a focus on markets coverage, analysing the wealth of Thomson Reuters data to identify facts and potential newsworthy items.
Unlike at other media organisations that have experimented with automation in financial journalism, the tool is not designed to write stories by itself. For Reuters, the future of automation in the newsroom is centred around tasks such as data-mining and gathering insights, which can then be presented to human journalists who can better contextualise and report on the findings.
The algorithm powering Lynx Insight was partly shaped by Reuters journalists, Padraic Cassidy, editor for news technology, Reuters, explained to Journalism.co.uk in an email.
"We surveyed our journalists and newsrooms to come up with lists of the most challenging questions they would like answered, in terms of gathering newsworthy information milestones, as well as what repetitive tasks they do now that could be handled by automation."
The help of the development team was then enlisted to create an experience with language templates and a simple interface that could be used by all journalists with only a few minutes of training.
"Templates for creating the sentences can be shared, vetted, favourited, allowing lots of individual beat-specific queries. The development team takes on the most complicated ones and returns it to journalists, who are the subject matter experts for final tweaking."
While Lynx Insight can return facts packaged as full sentences, such as "Shares of Johnson & Johnson (JNJ) are up 0.68 pct to $132.07 at 21:01 GMT," writing complete stories is not the focus for automated tools at Reuters.
Today's launch marks an important step towards a more automated workflow for data journalism in the newsroom.
"In the future we can much more easily create a graphic by forwarding on the data request for a sentence to our graphics interface, and allow reporters to 'see' what a milestone might look like visually, and possibly experiment with presentation," said Cassidy.
"We can also conceivably set an alert for an upcoming newsworthy milestone that can be shared with various teams’ calendars, to help more people prepare to contribute to a story.
"There are a variety of places where automation will help Reuters, including tagging and matching assets (pictures with text, video, graphics and text packaging) and news planning or calendaring. Audience engagement is also a good way to explore automation."
Free daily newsletter
- "Mass customisation" of neighbourhood data can help hyperlocal news become more sustainable
- World News Day: seven stand-out pieces of journalism in 2020
- Reuters launches AI-powered tool to speed up discovery of historic videos
- Tip: Make covid-19 data more accessible to readers
- Need data for your next story? This tool can get you the stats before the official figures are out