Nick Lockwood
I wrote "neet and tidy" and autocorrect changed "neet" to "need" instead of "neat", thereby masking my mistake and making it invisible to me
The state of autocorrect is embarrassingly bad. How is it that no attempt whatsoever has been made to make it understand sentence structure?
-
-
- View other replies
-
@nicklockwood Needs some machine learning :p -
@adamcnight@nicklockwood IIRC it uses machine learning in iOS 10. Makes it much worse
-
-
-
@nicklockwood I've seen cues that suggest they do try to understand *some* context ("were" vs "we're"). Just not very well. -
@radexp that's literally just checking if it's the first word in the sentence. It doesn't take any notice of what the preceding word was.
-
-
-
@nicklockwood has no attempt been made? I think it's just really hard and we can't currently do any better :( -
@H2CO3_iOS autocorrect kicks in after you type a word instead of at the end of a sentence. It's clearly not looking at the surrounding words
-
-
-
@nicklockwood I think iOS autocorrect has got an awful lot worse in the past year or two. -
@RichardBuckle I think that's true, but I don't think it ever used surrounding word context, so must be getting worse for some other reason.
-
-
@nicklockwood And why isn’t it editable yet? -
@nicklockwood seems like a perfect use for deep learning. That would be an awesome project to work on. -
@nicklockwood Seems like a case where low cost results are used. Wouldn't it be computationally expensive to do continuous syntax analysis?
Adam 'McNight'
Trick or Testut
Radek Pietruszewski
Árpád Goretity
Richard Buckle
Michael Tsai
Lee Higgins
Derek Klinge
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.