‘Changing the workflow of a newsroom is a major task. Integrating the values and specific style of a publication into the system is more of a challenge than operating the technology itself.’

– Maeve Hennessey

An autoscribe system has been going to school on Long Island, New York, helping publisher Newsday learn how to scale up local news coverage using data from the area’s education system. Both journalists and technologists have increased their know-how along the way.

The focus was on how repetitive, data-rich tasks could be offloaded to automation systems.

Catalyst funding came from by The LenFest Institute. The trial teamed Newsday with news automation pioneer AP. Lenfest says Newsday offered built-in digital resources and ‘a collaborative culture’ while AP brought a long-view in the space.

The trial turned out a number of pieces. This one shows portions of text prepared by the automated system outlined in yellow. IMAGE: Lenfest

Key takeaways reported by LenFest include:

  • Time – Journalists and technologists needed to find ways of working together. There were positive effects on the working culture, but it required extra time.
  • Customization by source – Each dataset has peculiarities. For example, protocols to retrieve one kind of data, such as weather, don’t automatically work with other kinds of data, like educational facts.
  • Augmentation – In some situations, automation frees-up time for journalists to do more complex tasks; in others, data sorting can reveal insights that become the basis for new stories. Neither will replace journalists.

OUR TAKE

  • The economics of automated journalism are a natural fit with dwindling cash for local news coverage. Systems can add fresh material targeted to very small geographies and give scarce journalistic time greater impact.
  • Other examples of automated local coverage in the U.S. are at Patch, and the Washington Post. The U.K. has the leading example of automated local news generation, at RADAR.
  • The Newsday experience surfaces an important caveat as more places contemplate autoscribe systems. Processes created for one dataset may not automatically apply to other data sets.

SEE RELATED STORY

LATEST