News Labs at French Assises of Journalism conference

Panel picture

Our own Basile Simon participated in a panel last week at the Assises of Journalism conference, in Metz, France.

The debate, organised by ObsWeb, was called: “Algortihms and predictions of the information: danger or opportunity for journalism?” It was moderated by Nicolas Becquet, multimedia journalist in Belgium, and involved Alice Antheaume, co-director of Sciences Po Paris journalism school, Benît Raphael, founder of Trendsboard, Emile Servan-Schreiber, CEO of Lumenogic, Guillaume Sire, professor at IFP Université Paris 2, as well as Basile.

The panelists discussed their practices and what the newsrooms do when it comes to involving the machine in the journalistic process.

Basile presented the Datastringer project and its goal: to help journalists consume data more easily, and to build a dataset monitoring tool tailored to their needs.

“I was talking with a team of BBC London journalists, showing them examples I worked on for this project, including a particular one monitoring some crime statistics in the UK. I just asked them: ‘What would be a good story about crime for you?’, to which they replied, and everybody would say the same thing: ‘A sharp decline or rise in crime.’ They said they would call ‘sharp’ a 25% difference compared to a yearly average. That’s all I needed to set up a Stringer for them.”

“Their stringer is going to talk with, the open-government platform for crime statistics, that publishes every arrest made in the UK with a bunch of informations: approximate GPS coordinates of the arrest, officers involved, category of the crime investigated, what happened next… The stringer will collate all these arrests in the Greater London area, and sort them by month and crime category (anti-social behaviour, arsony, bicycle theft…). Then, it will calculate averages, and check out the difference between the latest published data and these averages. If the difference is greater than plus or minus 25%, they will receive an email with the relevant numbers… and they know there’s a story to write.”

The debate also aimed to debunk pre-conceived ideas about algorithms, for example how newsrooms adapt their content to SEO criteria coming from Google, but also to show how journalists can be helped by these new tools in talking to their audiences, by the means of personalising the information for example.

Datastringer is available on Github and was built by Basile Simon and Clément Geiger.

Photograph at the top courtesy of ObsWeb on Flickr