News

The latest updates about Superdesk

BACK TO BLOG OVERVIEW

Connecting the Bots with Superdesk at NTB

NTB's elections bot
NTB's elections bot

Elections are important news events. That also makes them labour-intensive efforts at most news agencies, as reporters and editors have to sift through the continuous stream of data that pours in throughout the day and night of the election.

Norway’s national news agency NTB is making that process easier, and lightening the reporting burden for its editorial staff, with the use of a dedicated elections bot integrated with its Superdesk newsroom computer system.

Automating News Coverage  

Previously, two to three editors in the NTB newsroom were assigned to cover a general election. They would monitor the votes as they came in from each of Norway’s 19 counties, filing stories on the latest results. In September 2017, NTB assigned first-level reporting of the election to the elections bot, which sent its drafts to Superdesk just like any other reporter.

“Our main motivation was to be able to create bots ourselves and maintain that knowledge within NTB,” said Magnus Aaabech, project lead at NTB.

This wasn’t NTB’s first venture into AI-assisted reporting for its digital newsroom. A football bot was first out of the gate, back in May 2016. That project was successful, but had been developed with external resources. Going forward, the challenge for NTB's newsroom staff was to create more bots in-house. “We wanted to get people involved internally and also create a framework for something we could use in the future,” explained Aabech.

Connecting the bots to Superdesk

The elections bot was coded in JavaScript by Aabech and a fellow reporter. They then linked the bot to the Norwegian election directorate’s data feed via an API. When data matching certain criteria was found, the bot would create articles based on templates. These initial drafts were sent to a custom desk in the Superdesk news management system. An editor looked over these drafts, made any changes needed –  although sometimes none were – and sent the story on for publishing.

The bot submits its stories in Superdesk

“It was just like creating an article template. All the metadata was there when the file was ingested,” said Aabach. 

More than 80 stories were produced in this way.

Adding Digital Skills to the Newsroom

Though the process is still being refined, NTB considers the project a success on several counts. First, the bot accelerated and simplified the otherwise tedious task of going through large amounts of data on election night. Identifying the relevant numbers and doing some basic analysis allowed the journalists to focus on adding their own insights to an article, and telling a bigger story about the election results.

The election bot also enhanced the digital skill set of the editorial team. The journalists on the project wrote the code for the bot themselves and connected it to the election directorate API on one side and Superdesk as the headless CMS on the other.

Finally, the project is considered a success at NTB because the news agency’s customers published the stories the bot had written. The feedback from other journalists in the newsroom has also been positive. “They were impressed with the variations in wording given that it’s template-based. What we showed is that you can still have a variety of angles, so the stories don’t all sound the same,” said Aabech.

What’s the next bot project at NTB? Data-intensive subjects with national relevance – such as statistics about the economy – are one area of exploration. Another idea is to build bots inside Superdesk itself.

What can our newsroom software do for you? Get in touch with us to learn more about Superdesk.

BACK TO TOP

Latest Articles

Sign up for free to our monthly newsletter to receive Software updates and news about Superdesk

Demo request

Before you go

See Superdesk in action with a no-obligation demo for your organisation.

Schedule a demo