Robot Journalism


  1. Who is respon­si­ble for robot jour­na­lism?
  2. How is it used
  3. Why should news­pa­pers use auto­ma­ti­on?
  4. Why should news­pa­pers use auto­ma­ti­on?



Who is responsible for Robot Journalism?


The­re are ques­ti­ons over who claims aut­hor­ship of robot jour­na­lism, and with whom lia­bi­li­ty lies in cases of libel. The­re is not yet a defi­ni­ti­ve legal opi­ni­on on whe­ther the news orga­ni­sa­ti­on or the soft­ware com­pa­ny that pro­du­ces the robot jour­na­lism sys­tems have lia­bi­li­ty in cases of libel.

How does it work?


The tech­no­lo­gy behind robot jour­na­lism is cal­led Natu­ral Lan­guage Gene­ra­ti­on (NLG).

NLG pro­jects are for­med using tem­pla­tes and con­di­ti­ons. Tem­pla­tes are essen­ti­al­ly sen­ten­ces with gaps that are fil­led using data and lexi­ca­li­sa­ti­on algo­rithms. Con­di­ti­ons are the cir­cum­s­tan­ces that need to be met in order for the tem­pla­te to be uti­li­sed. The pro­cess hap­pens through a soft­ware engi­ne that arran­ges the tem­pla­tes into a deter­mi­ned order known as a “nar­ra­ti­ve”. The soft­ware that forms the basis of Retresco’s offe­rings is the rtr tex­ten­gi­ne. The pre­de­ter­mi­ned order that tem­pla­tes are arran­ged into is known as a ‘sto­ryp­lot’.

NLG is also able to use big data to spot news­wor­thy deve­lop­ments such as a sharp and unex­pec­ted increa­se in one value. This would trig­ger the pro­duc­tion of a spe­cial sto­ry high­lighting this shift. While a very basic use fills in gaps with data, more in-dep­th pro­jects ana­ly­se the data, make sen­se of it, and draws con­clu­si­ons. An examp­le of this is finan­ci­al reporting. Such a pro­ject invol­ves loo­king at raw data and inter­pre­ting not only what the various finan­ci­al events are but also ran­king them in order of impor­t­an­ce.

How is it used


Robot jour­na­lism is alre­ady pre­sent in major news­rooms and orga­ni­sa­ti­ons, par­ti­cu­lar­ly in the United Sta­tes. The­se inclu­de The Los Ange­les Times, For­bes, The New York Times, the Asso­cia­ted Press, and Pro­Pu­bli­ca.

In 2011, Qua­ke­Bot was laun­ched by The Los Ange­les Times. The sys­tem was con­nec­ted to the US Geo­lo­gi­cal Survey’s Ear­t­h­qua­ke Noti­fi­ca­ti­on Ser­vice. When the sys­tem recei­ved a noti­fi­ca­ti­on about an ear­t­h­qua­ke, it auto­ma­ti­cal­ly gene­ra­tes a news sto­ry with the sali­ent infor­ma­ti­on of time, loca­ti­on, and magnitu­de. The sto­ry is then pla­ced into the paper’s con­tent manage­ment sys­tem, whe­re it awaits appro­val from a human edi­tor. The sys­tem first gai­ned widespread reco­gni­ti­on in 2013 when it was the first to report that an ear­t­h­qua­ke of magnitu­de 4.4 had hit Sou­thern Cali­for­nia.

Why should newspapers use automation?


Algo­rithms are able to gene­ra­te quicker and fas­ter news that is poten­ti­al­ly less error-pro­ne than human-pro­du­ced con­tent.
Trans­la­ti­on is fair­ly easy to imple­ment, as sto­ries can be gene­ra­ted in mul­ti­ple lan­guages simul­ta­neous­ly from the same data­set.

NLG con­tent is limi­ted to a data pool. They are unab­le to con­duct inter­views, draws lines of cau­sa­li­ty, or be able to pro­vi­de much—or any—external con­text.