Turing’s Test 113
As with all kinds of data projects, cleaning and preparing the data is an
immensely important activity, but once organised, it is the selection and
use of templates that allows throughput, the transformation of those
data into some form of narrative. The primary elements of such throughput via templates are the use of branches – as indicated previously with
a simple example dealing with increasing or decreasing share prices –
and synonyms, allowing variations on similar structures when converting data into stories. After a data structure has been defined, templates
are constructed using Boolean true/false formulas (which, if commonly
employed, can be further converted into templates nestled into other
branching templates).
Between 2007 and 2013, automated or robot journalism had made
huge developments, but the parameters of such work were extremely
limited: while services such as Automated Insights and Narrative Science were clearly capable of producing a greater number of pieces of
content than the combined workforce of media organisations, many of
these were very simple presentations of data such as charts or league
tables and even when there were narrative stories, these were confined
to extremely clearly defined, highly structured categories such as sport
and finance where data could be presented in a non-messy format. More
surprising, however, was an announcement in late 2017 that automated
journalism could be used to write local news. In December 2017, AP
announced that it was setting up a trial of a publishing service in the
UK and Ireland called RADAR (Reporters and Data and Robots) in
conjunction with Urbs Media and 14 local publishing groups (including
Johnston Press, Newsquest, and Trinity Mirror), having made its first editorial hires and launched a pilot in November.
59
The service, launched
with investment from the €150 million Google’s Digital News Initiative
and tested out hyperlocal variants of stories focussing on trends in birth
registrations, cancelled operations, a breakdown of social mobility and
life chances for disadvantaged children, and localised traffic data. These
stories, published by 20 titles, were the first automated local news stories
to be published by established brands in the world, and all drew heavily on data drawn from the Office of National Statistics, the NHS, the
Social Mobility Commission, and Department of Transport. One such
story, by Tom Matthews and Ralph Blackburn, identified as a “Radar
Data Reporter”, was published as follows in the Croydon Advertiser on
4 December, 2017:
Seven potentially life-saving operations were cancelled in
Croydon in October
Latest health data have revealed that the body which runs Croydon’s
hospitals was one of 40 trusts in England to cancel at least one important procedure.
As with all kinds of data projects, cleaning and preparing the data is an
immensely important activity, but once organised, it is the selection and
use of templates that allows throughput, the transformation of those
data into some form of narrative. The primary elements of such throughput via templates are the use of branches – as indicated previously with
a simple example dealing with increasing or decreasing share prices –
and synonyms, allowing variations on similar structures when converting data into stories. After a data structure has been defined, templates
are constructed using Boolean true/false formulas (which, if commonly
employed, can be further converted into templates nestled into other
branching templates).
Between 2007 and 2013, automated or robot journalism had made
huge developments, but the parameters of such work were extremely
limited: while services such as Automated Insights and Narrative Science were clearly capable of producing a greater number of pieces of
content than the combined workforce of media organisations, many of
these were very simple presentations of data such as charts or league
tables and even when there were narrative stories, these were confined
to extremely clearly defined, highly structured categories such as sport
and finance where data could be presented in a non-messy format. More
surprising, however, was an announcement in late 2017 that automated
journalism could be used to write local news. In December 2017, AP
announced that it was setting up a trial of a publishing service in the
UK and Ireland called RADAR (Reporters and Data and Robots) in
conjunction with Urbs Media and 14 local publishing groups (including
Johnston Press, Newsquest, and Trinity Mirror), having made its first editorial hires and launched a pilot in November.
59
The service, launched
with investment from the €150 million Google’s Digital News Initiative
and tested out hyperlocal variants of stories focussing on trends in birth
registrations, cancelled operations, a breakdown of social mobility and
life chances for disadvantaged children, and localised traffic data. These
stories, published by 20 titles, were the first automated local news stories
to be published by established brands in the world, and all drew heavily on data drawn from the Office of National Statistics, the NHS, the
Social Mobility Commission, and Department of Transport. One such
story, by Tom Matthews and Ralph Blackburn, identified as a “Radar
Data Reporter”, was published as follows in the Croydon Advertiser on
4 December, 2017:
Seven potentially life-saving operations were cancelled in
Croydon in October
Latest health data have revealed that the body which runs Croydon’s
hospitals was one of 40 trusts in England to cancel at least one important procedure.