How PA and Urbs Media will use robots to strengthen local news, rather than devalue it

Reporters and Data and Robots (Radar) is a major service that launches next month and aims to supply the local news media with up to 1,000 stories a day, using automation to probe such areas as the performance of local hospitals, schools and police forces.

The initiative, founded by the Press Association and data news startup Urbs Media, has the potential to transform local news coverage by placing a fresh emphasis on evidence-based public service journalism and providing a compelling alternative to some of the press-released material processed by hard-pressed regional reporters. Traffic on local news websites can be driven by more than the D-list celebrity opening of a new fried chicken outlet.

The beauty of the Radar service is that the same thematic story, (the response times of ambulance services for example), can be repackaged for hundreds of outlets according to whether local operators are performing above or below national averages. The new service is being funded with a €706,000 (£639,300) grant from Google’s Digital News Initiative.

Gary Rogers, co-founder and editor-in-chief of Urbs Media, says that by deploying Natural Language Generation (NLG) software a Radar writer can compile a single template that will conigure the same story in multiple variations, determined by the data. So a single article can be tailored to cover each of 200 local authority regions and supplied to appropriate media.

Rogers talks with excitement of the possibilities of scouring open data sources for hyperlocal stories. Crime patterns, for example, can be compared in minute detail across 35,000 Lower Layer Super Output Areas (LSOAs) in the United Kingdom, each one equivalent to a neighbourhood of around 3,000 people. He has compiled a “heat map” of bicycle thefts, examining every reported offence over 12 months.

“Using reported crime data, I could produce 100 stories, one for nearly every town in Britain, identifying the worst place to leave your bike in any town,” he says. “That’s the heavy lifting we want to do.” Radar aims to accompany many of its data-driven stories with auto-generated graphics for local news sites.

He is currently looking at the numbers of foreign workers registering their National Insurance numbers in each local authority area, and at whether they are from countries inside or outside the European Union. “Those interesting patterns are all there in open data if you want to spend enough time with it and you have software to look at 200 or 300 different areas,” says Rogers. “That’s all publicly available and released on a regular basis, so you can get a picture of the changing workforce across the UK.”

Rogers believes that stories based on open and official material that can be cross-checked by readers are especially powerful in a time when the news media is being damaged by deliberate fake stories from shadowy sources. “We have always linked to our data sources and shown our workings,” he says. “I would hope we will be able to continue with that and show people stories that they can check if they have any doubt. People have an appetite for news that they can trust and which is verifiable.”

Another of his current probes involves scanning Department of Transport data for changing levels of ownership of diesel vehicles in each local authority area. “In many parts of the country diesel car ownership is till going up and up, even since the Volkswagen emissions scandal.”

The common theme with these stories is that journalism isn’t giving way to automatons, despite Radar's witty acronym. When Urbs Media was founded in September 2014 its strap line was “written by a human produced by a robot”.

Rogers points out that “it still takes journalistic interest to work out the story”. Until now, the startup has been using its Urbs London platform to showcase the value of data journalism in breaking down stories across the 32 London boroughs. It recently produced a microscopic picture of how childhood obesity varies between electoral wards.

But by partnering with the Press Association, which was originally founded in 1868 by an alliance of newspaper owners and today is known as the national news agency for the UK and Ireland, Urbs Media can take its data-driven content industry wide (to bloggers and small independent publishers, which might buy content through micro-payments, as well as big local news groups).

“We want to see stories spread very widely and the PA has the reach to achieve that, which is why it is a good partner for us,” says Rogers. Crucially for such a young news startup, PA can offer the specialist reporting resources which will add weight and context to the stories it unearths.

“If I’m working with health data set, I can work with the health editor of PA and get their insight to guide my view of the numbers and build some national and political context into the story,” says Rogers. “The trust and reputation of the PA is also important in establishing Radar as a strong source of news and the distribution mechanism we will build with the PA to get the content to people is absolutely vital. We can do these stories but unless we connect them with the markets they need to be in there’s no point.”

Urbs Media was runner-up in the international Startups for News 2017 competition organised by the Global News Network. It has small offices in London and Bristol but is likely to base a fivestrong team at PA’s headquarters in London’s Victoria in the coming months before Radar goes live in 2018, PA’s 150th anniversary year.

Before co-founding Urbs Media, Rogers worked in senior editorial management at both the BBC and ITN. Alan Renwick, co-founder and chief executive of Urbs Media, is a former head of strategy at major local news publisher Local World (which was acquired by Trinity Mirror in 2015).

Radar becomes part of the growing ecology in data journalism taking root in UK newsrooms. The Times, the Guardian and the Telegraph are among national titles to have invested in specialist data teams. The Local is a project being run from the Bureau of Investigative Journalism to produce data-driven investigative pieces for local news media, while DMINR is a fact-checking initiative using machine learning being hosted by City University in London and funded by another Google DNI grant.

Although Rogers would like Radar to generate front page news, the service’s aim is to maintain a high volume of well-sourced content that helps local media to function.

“We don’t want to be an organisation which produces a really big data story once a month, there are other organisations in the market doing that and all power to their elbow. What we want is a highly-productive, scaled operation of data stories - that’s what our measure of success will be.” Radar hopes to supply titles with around five stories a day, tailored to their patch.

Local journalists might be alarmed that such a story machine might be seen as a way to make further staff cuts in a sector that has already radically reduced its headcount. However, Renwick emphasises that Radar is there to make it easier for locals to publish quality content. “This is going to be disproportionately helpful for smaller titles,” he says. “If we can give three or four stories a day for each of these titles it’s a really good bedrock.”

Pete Clifton, editor-in-chief of the PA, is convinced that Radar will offer a “cost-effective way” of providing “incisive local stories, enabling audiences to hold democratic bodies to account”.

If Radar succeeds, Renwick believes the model can be rolled out internationally. “We have already had discussions with some of the bigger markets overseas,” he says. “Clearly if this works in the UK and Ireland it will work anywhere.”

Ian Burrell's column, The News Business, is published on The Drum each Thursday. Follow Ian on Twitter @iburrell