But why Jinx?
Word choice and grammar, alrighty. Context? Not so much: Three-Week Abortion Day In Pictures.
Edit: Can someone rig up a bot to reply to the posted headlines/articles?
Monday Night’s Sex “In The Game,” Says Gay & Child Of Chicago Music
This stuff is important! Why didn’t I know about this sooner?
I don’t get it, really.
Non-artificial intelligence hates it!
A slightly (slightly) less ridiculous idea has been in the works for a while now with The Grid (linking to FAQ b/c their home page isn’t worth the effort to understand).
On second thoughts. I think The Grid may in fact just be a scam (Irresistible proposition? Founding member invites? Non-refundable payments? Why on earth would one think that??)
Ah god that’s so much better! Thank you.
Off in the corner, rocking, sucking his thumb, and very certainly not being jealous.
Some websites are in danger of becoming Click-o-trons.
It’s ok.
There’s something really pleasing to me about reading this stuff. Like, by breaking clickbait down into a slush it becomes easier to digest. Without any real context all that’s left is the pure essence, more easily studied. Examples:
A man was arrested after reportedly spending hours on the streets of an online video for his latest film “How to fight the British Music.” He was killed by an in-flight officer on last night’s episode
Difficulty distinguishing between real and fantasy violence.
A black and white girl is accused…
Race is relevant if the story is criminal in nature.
Really not the first. The AP has had a robot writing articles for a while now. The system is basically used for stock market and sports reporting- Articles which are mostly just straight data with a few adjectives thrown in.
There are also books produced using a similar process- One guy has “authored” over 1,000,000 of them.
@responderbot: abortion, cats, feces
@yeahnobot: yesorno?
Philip M. Parker, Narative Science, et alia are primarily using different methods – sophisticated analysis + sophisticated templates, primarily – to create readable narrative from data. They’re doing pretty well, and making book. So to speak.
Click-o-Tron isn’t trying to produce a linear narrative – it’s using some sophisticated “dumb” methods to generate text that weaves back-and-forth from source material that it has been told are related… but aren’t all that related. Lingustically they are related, but at a larger level of atomicity the semantics break down, and this fun stuff is the result. It’s a smarter n-gram chain (that’s not a dig, and it might even be wrong, since I haven’t studied recurrent neural networks at all, yet.).
On-going is the third year of mad-science project known as NaNoGenMo - Nation Novel Generation Month - spend the month of November writing code to write 50,000 words.
It’s a “hopeless” goal, in that academics and professionals have spent literally decades on specific projects that have a hard time making a short story (but a collection of short stories counts!).
Full disclosure - As in previous years, I’m a participant - see https://github.com/dariusk/NaNoGenMo-2015/issues/14 and https://github.com/dariusk/NaNoGenMo-2015/issues/162
Does this mean all those sites have been using human beings to write those headlines all this time?! There ought to be a law.
