Half a croissant, on a plate, with a sign in front of it saying '50c'
h a l f b a k e r y
It's the thought that counts.

idea: add, search, annotate, link, view, overview, recent, by name, random

meta: news, help, about, links, report a problem

account: browse anonymously, or get an account and write.

user:
pass:
register,


                                       

Please log in.
Before you can vote, you need to register. Please log in or create an account.

Hyperpredictive text

Attempts to complete your entire communication.
  (+10)(+10)
(+10)
  [vote for,
against]

You are probably aware of an autocomplete function in at least one context. Usually, these try to finish your word for you on the basis of the first few characters. This is common on mobile phones. In the case of Google's predictive search, it uses the preceeding words in the text-box, and relies on everyone searching for the same thing.

A Markov text generator is a system which, given a set of text, generates a stream of similar text based on the probability of word combinations.

I propose to marry these two concepts. When typing something, say an article, the system attempts to finish the entire article based on what you've done so far, and derived from the frequencies of combinations in articles you've previously written. You'd be at liberty to accept as much of this as you like - which might be several sentences, with a well-trained system.
Since most people write articles in several different contexts, it would be desirable to be able to tag articles, so the system could use different weightings for each. Over time it could 'learn' not only your general turn-of-phrase, but also when you say it, so you wouldn't have to.

Loris, Dec 28 2010

Markov Chain Algorithms http://www.in-vacua.com/markov_text.html
(A not very technical explanation) [Loris, Dec 28 2010]

Prediction by partial matching http://en.wikipedia...by_partial_matching
[Spacecoyote, Dec 30 2010]

Dynamic Markov Compression http://en.wikipedia..._Markov_compression
[Spacecoyote, Dec 30 2010]

Funny Results http://thedailywtf....c-WTF-No-Quack.aspx
...from a neural network [Spacecoyote, Dec 30 2010]

A.R.T.H.U.R.'s Anthology of English Poetry by Laurence Lerner http://www.poemhunt...-of-english-poetry/
Not Markov as far as I know, but generated by selective cut'n'pasting. A.R.T.H.U.R - Automatic Record Tabulator but Heuristically Unreliable Reasoner [Dub, Dec 30 2010]

A.R.T.H.U.R. http://lists.digita...10-June/001321.html
[Dub, Dec 30 2010]

Yes, this should work flawlessly http://damnyouautocorrect.com
[BunsenHoneydew, Jan 02 2011]

ChatterBot ChatterBot
[theircompetitor, Jan 02 2011]

Emacs "Dissociated Press" https://gnoetrydail...ted-press-in-emacs/
Baked since 1972. In emacs, command "M-x dissociated-press" in a buffer full of text. [rmutt, Jan 03 2011]

[link]






       Drive-by bunning.
nineteenthly, Dec 28 2010
  

       This could possibly be done in a top-down fashion: given writing examples a machine could write an article, given a very very detailed outline: sort of a text compiler.   

       But I doubt it could be done in a linear fashion, which the Title suggests, where you start dictating an article and the machine finishes it for you, given that the machine has no clue what point you're trying to make.   

       "Uncle Tobias we kept in a bucket..."
FlyingToaster, Dec 28 2010
  

       I like the idea.   

       I just worry about the distraction.
mofosyne, Dec 28 2010
  

       ... stay tuned another couple of years (or decades depending how it goes) - I'm baking something just like it.

What you are basically describing is form of AI. Imagine injecting this idea into a to a human brain: "Uncle Tobias we kept in a bucket...." .. the person would either complete the story for you best as they could, or they would get overwhelmed all the ambiguous completions and give up. The difference between those two results depends on the persons "creativity quotient" ... the more creative person will be willing to take risks and just "output" whatever ridiculous completion comes to mind. The less creative person will be more careful and just say "I have no idea" and stop at that.
So basically there would need to be some kind of a "creativity" slider along with this input form. Low creativity gives you brief results that are well accepted and known, high creativity results would be lengthy and creative.
ixnaum, Dec 29 2010
  

       Probably useless, but would kick out funny results. I look forward to the results of [ixnaum]'s research.
wagster, Dec 29 2010
  

       Swipe aaahhhhhh
zeno, Dec 30 2010
  

       This is a thoroughly excellent idea; if baked, I wouldn't care whether the output were meaningful; in fact, generation of literate, creative rubbish would probably be the commonest use; definite morning goods
H K Boris, Jan 01 2011
  

       <clippy>   

       It looks like you're trying to write a resignation letter. Would you like me to finish it for you?   

       </clippy>
Cedar Park, Jan 03 2011
  

       [+] Sort of like Autotune, but for writing. You could just bang away at the keyboard, and it would Autotext in real time, making you sound great (comparable to Cher singing 'Believe.'). Also, that way, if you were accused of plagiarism, you could blame the Autotexter.
sqeaketh the wheel, Jan 03 2011
  

       It might have a niche application in the treatment of writer's block. For people who have particular horror of the blank page ("Writing's easy. You put a blank sheet of paper into the typewriter, then stare at it until drops of blood form on your forehead") it's sometimes recommended to type anything -- anything at all, no matter how bad, just so the page isn't blank anymore. That gets you over the hump, and then you can revise your way towards the final product. For some of us, revision, no mater how extensive, is infinitely easier than writing the text in the first place.   

       [rmutt] emacs doesn't count as baked, because if it did *everything* would be baked.
mouseposture, Jan 04 2011
  
      
[annotate]
  


 

back: main index

business  computer  culture  fashion  food  halfbakery  home  other  product  public  science  sport  vehicle