Newsworthy’s automated alchemy, transmuting data into story leads
A look at the journalism tool meant to give small newsrooms a first step toward using big data. The tool earned a place on the Data Journalism Awards 2018 shortlist for its innovation in data journalism.
‘Local newsrooms often lack the skills and resources to work with data-driven reporting. Even just opening an Excel sheet is frightening to many’, wrote J++ Stockholm founder Jens Finnäs.
To address this fear Finnäs, Leo Wallentin and their team at Journalism++ Stockholm developed the Swedish-based journalism service, Newsworthy. Newsworthy generates story leads for newsrooms by analysing large amounts of information in a hands-off approach to data-driven reporting.
How does it work?
First, Newsworthy requires a large data set to pull from. This process can be applied to any data that can be described as a statistical series over time.
Using calibrated statistical models, Newsworthy then scours the accumulated data and picks out overarching trends or anomalies that stand out from the norm, whether that be changing trends in weather reports or a sudden spike in local crime.
Once it has detected something noteworthy, it then generates a lead, a short text, a chart and a spreadsheet describing the findings. This report provides reporters with quick local content or the foundation on which to build a broader story.
‘With minimum effort we are able to generate thousands of local [news leads]’, wrote Finnäs.
Benefits and challenges of automation?
Calibration — Newsworthy offers a particular benefit to newsrooms with limited personnel resources. By automating data collection, analysis, standard editorial decisions, text generation, distribution, etc., Newsworthy saves reporter’s time and energy. However, this automation requires a substantial initial commitment as the algorithms must be attuned for each new data set, but becomes more valuable the more it is used.
‘We have [learned] that there is not one magical algorithm that can be applied to any data and output meaningful news’, wrote Finnäs. ‘We have to define newsworthiness with different criteria in weather data than, say, housing prices. Context matters.’
Newsworthiness — While Newsworthy’s algorithm must be adapted to determine what qualifies as notable for each new data set, they all work on a binary system, picking out only what passes the pre-set threshold of newsworthiness. However, Finnäs hopes to make the system more qualitative in the future.
‘I’m hoping we’ll be able to move towards a more granular assessment. How newsworthy is a given time series on a scale from, say, 0–10?’ wrote Finnäs.
Leads not stories — Newsworthy emphasises that their findings are leads to stories, not a final product themselves. The data can be used to generate cheap, local content with little reporter intervention or be the basis for a larger story supplemented by traditional reporting.
As an example of Newsworthy’s potential, Finnäs points to a Swedish public radio reporter using Newsworthy to uncover that the city of Östersund was buying social houses and apartments based on a peak in real estate sales.
‘I like this example because the reporter actually went out to find the explanation for the anomaly — and found a story’, wrote Finnäs. ‘[In] practice we often see that the stories are just published as such.’
Business model — branching out while staying local
Currently, when you visit Newsworthy’s website, you can subscribe to a free version of the service which provides weekly updates on weather and monthly reports on unemployment and housing for countries across Europe. According to Finnäs, they will soon be offering Swedish newsrooms a monthly subscription of 150 euros per newsroom for access to all of their data coverage.
The original plan for Newsworthy was to build revenue through this traditional model, charging newsrooms or a group of newsrooms on a subscription basis. The economic reality facing many local newsrooms puts the feasibility of this model into question.
‘I don’t think we’ll be able to build a sustainable economy only on traditional newsrooms’, wrote Finnäs.
According to him, Newsworthy is now pursuing non-journalistic clients to augment their user-base. One of these unconventional clients is the Swedish teacher’s union, Lärarförbundet, who use Newsworthy to generate locally-angled press releases and reports on data relevant to the organization.
‘ … For now (and in the near future) we will be paying the bills with revenue from partnerships with non-journalistic clients, such as Lärarförbundet’, wrote Finnäs. ‘For them, we are more of a sub-contractor or content provider, than a news service.’
If you want to see what other projects were included in the Data Journalism Awards 2018 shortlist you can click here and to see the winners click here. If you are interested in learning more about what J++ is up to click here.
