Press Association is using artificial intelligence to help local news organisations produce more data-driven stories

The Press Association (PA) is taking a new approach to supporting local journalism over the next year, through the launch of its new service, 'Reporters and Data and Robots' (RADAR), which will use automation to create a stream of local stories for hundreds of media outlets.

PA and its partner, data-driven news start-up Urbs Media, received €706,000 (£621,000) from Google's Digital News Initiative (DNI) in July to develop the new service, one of the largest grants allocated so far through the Innovation Fund.

Their aim is for RADAR to enable local news outlets to publish more stories on issues that affect their readers directly, on subjects such as local crime, education and health.

You'll get stories that are worth doing more on that you can further investigate – it's a foundation that you can do more withPete Clifton, Press Association

Pete Clifton, editor-in-chief of the Press Association, explained that RADAR will use automation and artificial intelligence to create story templates by analysing national open databases, at a volume that would be impossible to achieve manually.

"At PA, as a national news agency, we might write about the big trends or figures, but within lots of these data sets, if you burrow into them enough, there'll be a line in the data that relates to almost every postcode in the country," he said, speaking at WAN-IFRA's International Newsroom Summit in London today (9 November).

The new service, which will begin a six-month trial next year, will create up to 30,000 localised stories each month from open data sets, from institutions like government departments, local authorities and NHS Trusts.

Urbs Media is in the process of hiring six data journalists to work with the software, who will bring their editorial expertise to the process, and add relevant keywords and phrases to stories that will help with the automation.

"RADAR is about us harnessing publicly available data with the right mix of journalists and automation – it's a level of journalism that we could never do with the amount of reporters that we have," he said.

"I am not expecting that anybody is going to say 'great let's publish it', because if you read 100 of [these automated stories] you'd realise there was a rhythm to them.

"But if you work in a local newsroom, you'll get stories that are worth doing more on, that you can further investigate – it's a foundation that you can do more with."

Natural language generation (NLG) software will be used to produce multiple versions of stories, to "scale up the mass localisation of news content".

Twelve of PA's existing customers are going to test-run RADAR this year, but when the trial begins running in 2018, the service will be free for the first few months, so that PA can gather feedback and see how it is performing.

"The second year of the programme will be about using the same techniques to start automating basic graphics and animated videos to go within pieces of copy we send out," Clifton said.

"I still think they will need more work to turn them into something that can really be used, but they will serve as building blocks to make it closer to being something you can publish."

The DNI grant will also be used to boost the news agency’s distribution platforms, to help more people find and use the content.

Free daily newsletter

If you like our news and feature articles, you can sign up to receive our free daily (Mon-Fri) email newsletter (mobile friendly).