While there is a lot of discussion in media circles about the potential of automated content, real world examples of bots creating and publishing editorial are still fairly thin on the ground. Bringing bots into the news room can, after all, create many issues for media company owners. From finding the right way to deploy them through to training journalists to optimise their output, the journey to automation can be a tricky one.
There are however a few very good examples of best practice in automated content, and one groundbreaking collaboration that has gone from strength to strength in recent years is the work between Swedish publisher MittMedia and automation experts United Robots.
At DIS 2019 MittMedia’s Chief Digital Officer, Robin Govik, will explain how the collaboration came about and how it is transforming the media company’s sports and local news output. Here he explain about his role at MittMedia and offers his take on how bots and journalists might complement each other in the future.
***Take advantage of our early bird offer for DIS 2019, taking place on 25 to 26 March in Berlin. By signing up for DIS 2019 by January 31st, you will save €400 on final delegate rates. Sign up here***
Tell me about your career so far - what was your journey to MittMedia?
I've been a journalist all my life. But always with a big interest in technology marketing. For 12 years I worked for Schibsted with their publication Aftonbladet in Stockholm – Sweden's biggest daily. Aftonbladet was a pioneer online in Scandinavia. When they launched an online edition in 1994 no one understood that that would be their main channel just ten years later. When my wife and daughters decided to move out to the countryside MittMedia was the perfect fit. MittMedia is a publisher of local and regional news brands and has been very strategic in achieving a digital-first position. To be able to stay relevant you must adapt to new user behavior and MittMedia has been very focused on the users.
What is your main role at MittMedia? How are you using data, and what tech is being used to support and develop its role within the company?
As Chief Digital Officer I'm responsible for our digital development and transformation. In my organisation, there's a lot of different skills: Journalists, product managers, software developers, data scientist, content developers. Just to name a few.
Data is at the center of our digital ecosystem. That means that we both have the technology to collect the data and an organisation that can work with the data.
We have developed a data platform that we call "Soldr". In Soldr we collect and store three data points: content, users and interactions.
Mittmedia’s data team is organised around the different types of data used as the backbone of the digital organisation. This means that we have competences within the group to ensure that the team autonomously can work within the scope of collecting, processing and extracting user content/interaction and advertising data. To be able to do this we have devops, machine learning experts and data scientists working together.
For example, how is audience data used to drive advertising and subscription revenue?
Audience data is deeply integrated into our products and processes. As an example: we use audience data to present a personalised selection of what stories to display to the users. In the same way, we also use audience data to build target groups for advertisers. Of course, we use audience data to build insights that editors and product managers can use to make better content and better products.
You are also chair of United Robots - explain what the company does and how did your role in that company come about?
United Robots uses Artificial Intelligence (AI) and Natural Language Generation (NLG) to automatically produce publishable news texts from large data sets, including sports, property sales, and local business registrations.
MittMedia uses automated content in order to make a better news product. Since we cover such a big geographical area with local news we face a lot of challenges. We need to use our editorial resources in a smart way. This means that we have local reporters in almost all of our communities covering local events. But the demand for local news content is bigger than what we can produce with human journalists. That's why we complement the news feed with automated content.
In 2015 we started to experiment with automated content with MittMedia’s own developers. But we came, pretty quick, to the conclusion that it makes much more sense to do this together with others that can scale the product. So at the beginning of 2016 we founded the company, United Robots, together with two partners, to develop a platform for automated content.
What inspired the recent collaboration between the two companies in the creation of sports content? And what were the key challenges you had to overcome?
We could see in our data that our digital subscribers were really interested in sports. And it was a broad interest. United Robots then developed a Sports bot. The second match results are reported the Sports Bot automatically and immediately creates a publishable summary. The texts can either be published in the condition they are or serve as a draft for a reporter to work on. This provides several advantages. The cost per published text will be dramatically lower than if produced manually. And the process is ofcourse immeasurably faster.
Do you think that MittMedia will extend its use of bots to create content?
Yes, I’m certain that we will both make the existing automations better but also extend the content into new areas. Today we have a good collaboration between our highly skilled journalists and the automations.
What are the other areas that you think that bots will create content? And how do you think this will impact on the role of journalists within news organisations?
It’s easy to get carried away and start making robots that are just fun to develop instead of solving a user need. We can see today that one of the most important and most well read topics for our subscribers is crime. So it would make sense exploring an automation that covers everyday crime, like burglary, traffic crime and even shoplifting, in our communities. But to be able to build an automation we must have high quality data and that’s lacking then it comes to the police reports. Hopefully we can persuade the government to open up better data feeds.
With United Robots doing the legwork on basic editorial coverage, our reporters are freed up to work on qualitative journalism.
Where do you think bot content creation will finally lead us? Will it for example change the creation and monitoring of social media content? Are we going to see more bot created video?
There’s already bots that can write summaries of news articles – even with different angles. Like with a left wing bias or a right wing bias. There are bots interpreting videos.
The possibilities with automation of content is, of course, endless. I believe this is the future. Man and machine together to make great content.
That means that some journalists will keep doing investigative journalism, asking the tough questions to the local politicians and writing those stories that just gives you goosebumps.
At the same time, other journalists will focus on being the journalistic logic in autonomous systems. They will develop algorithms that make those personalised selections. They will teach the robots how to write an article. I hope that journalists will be in charge of the robot revolution, and embrace the new technology, when it comes to editorial content. Otherwise it’s not journalism.
***Take advantage of our early bird offer for DIS 2019, taking place on 25 to 26 March in Berlin. By signing up for DIS 2019 by January 31st, you will save €400 on final delegate rates. Sign up here***