‘The organization undertook its biggest test of machine learning-generated journalism to date, publishing nearly 700 individual news reports (649 in English, 40 in Welsh) throughout the night.’
An autoscribe pumped out BBC News election stories tailored to each UK constituency, TechHQ reports. Only one riding missed getting customized results because their counting wasn’t finished.
The scale and speed of the reports would never have been possible with human reporters, says the BBC News editor responsible for the project.
The UK has 650 constituencies.
The articles had no analysis and no quotes. Story structures were pre-set in a template. The autoscribe plugged in the relevant data ‘in the form of comprehensive sentences.’ BBC News editors vetted the automated stories before publication.
BBC used a natural language processing system similar to others at work for RADAR in the UK and the Associated Press. The systems are best suited to data-rich stories.
Machines wrote the BBC’s UK election results articles
TechHQ | December 13, 2019
SEE RELATED STORY
- General election 2019: Semi-automation makes it a night of 689 stories – BBC News Labs blog | ‘In total the team published approximately 100,000 words in 10 hours…’