News for hackers. Stuff that matters.
Re:IBM is not doing "cognitive computing" with Watson (2016) by emodendroket on Monday April 30, 2018 @04:04PM (#16961760)
I got a two-in-one Windows notebook and I was really, really impressed with the handwriting recognition. It can even read my cursive with like 95% accuracy, which I never imagined. That said, it's still slower than typing.
Re:IBM is not doing "cognitive computing" with Watson (2016) by mojomark on Monday April 30, 2018 @04:27PM (#16961960)
> ...That said, it's still slower than typing.

Finger-on-nose! It seems to be a question of utility and efficiency rather than primarily an emotional response to effectiveness of a given UI.

I type emails because it's faster. I handwrite class notes because I often use freeform groupings and diagrams that would make using a word processor painful.

Re:IBM is not doing "cognitive computing" with Watson (2016) by emodendroket on Monday April 30, 2018 @04:31PM (#16962009)
On the other hand, learning to type took some effort and I see many people who do not know how to properly type. The case for handwriting might have looked better if you assumed people would not all want to become typists.
Re:IBM is not doing "cognitive computing" with Watson (2016) by snowwrestler on Monday April 30, 2018 @11:44PM (#16964448)
Back in the early 2000s, the hybrid models (the ones with a keyboard) became the most popular Windows tablet products. A keyboard is a pretty convenient way to work around the shortcomings of handwriting recognition.

Most owners eventually used them like laptops unless illustrating, or operating in a sort of "personal presentation mode," where folding the screen flat was convenient. For most people, handwriting recognition was a cool trick that didn't get used much (as I suspect it was for you, eventually).

The utility of a handwriting recognition system is ultimately constrained by the method of correcting mistakes. Even a 99% accuracy means that on average you'll be correcting about one word per paragraph, which is enough to feel like a common task. If the only method of correction is via the pen, then the options are basically to select from a set of guesses the system gives you (if it does that), or rewrite the word until it gets recognized.

But if a product has a keyboard, then you can just use the keyboard for corrections. But if you're going to use the keyboard for corrections, why not just use it for writing in the first place...

Re:IBM is not doing "cognitive computing" with Watson (2016) by emodendroket on Tuesday May 01, 2018 @09:01AM (#16966651)
That's all true, but for a phone the writing input might be worth considering. That might be too optimistic though.
Re:IBM is not doing "cognitive computing" with Watson (2016) by snaky on Monday April 30, 2018 @05:27PM (#16962555)
> it's still slower than typing

In alphabetic languages. What about Chinese, or Japanese?

Re:IBM is not doing "cognitive computing" with Watson (2016) by DonaldFisk on Monday April 30, 2018 @06:25PM (#16963014)
There is software that allows you to draw the characters on a touch screen, but it's still slower than typing and most people either type the words phonetically (e.g. in Pinyin) or use something like Cangjie which is based on the shape of character components. Shape-based methods are harder to learn but can result in faster input.

For Japanese there are keyboard-based romaji and kana based input methods.

The correct Hanzi/Kanji is either inferred from context or, as a back-up, selected by the user.

Re:IBM is not doing "cognitive computing" with Watson (2016) by emodendroket on Monday April 30, 2018 @07:11PM (#16963318)
Almost nobody uses kana input in Japanese. There are specialized systems for newspaper editors, stenographers, etc., that are just as fast as English-language equivalents, but learning them is too difficult for most people to bother.
Re:IBM is not doing "cognitive computing" with Watson (2016) by opencl on Monday April 30, 2018 @07:52PM (#16963551)
Huh? The standard Japanese keyboard layout is Hiragana-based, and every touchscreen Japanese keyboard I've used defaults to 9key kana input. Do people really normally use the romanji IME instead?
Re:IBM is not doing "cognitive computing" with Watson (2016) by emodendroket on Monday April 30, 2018 @08:25PM (#16963711)
Yes. Despite the fact that kana are printed on the keys, pretty much no Japanese person who is not very elderly uses kana input. They use the romaji (no N) input instead, and that's the default setting for keyboards on Japanese PCs.

You are right that the ten-key method is more common on cell phones (and older feature phones had kana assigned to number keys), although romaji input is also available.

Re:IBM is not doing "cognitive computing" with Watson (2016) by emodendroket on Monday April 30, 2018 @07:09PM (#16963308)
Typing in Japanese is about half as fast as English if you're good. Still better than handwriting.
Re:IBM is not doing "cognitive computing" with Watson (2016) by GolDDranks on Tuesday May 01, 2018 @04:13AM (#16965338)
On the other hand, typing on a smartphone is faster in Japanese. (If you use the 10-key input method.)
Re:IBM is not doing "cognitive computing" with Watson (2016) by emodendroket on Tuesday May 01, 2018 @09:20AM (#16966788)
Not sure about that. I guess it depends on how good the prediction is for your text, but in general the big problem is that the step where you select what character you actually meant takes time.