this is amazing and will change life forever: live text translation via augmented reality

watmm is a place i come to visit when i need to be reminded how jaded we as westerners have become

The tech is cool for sure. This particular application of the tech is not necessarily life-changing, but the tech is cool as fuck.

  On 12/18/2010 at 9:23 PM, gaarg said:

There are other possibilities of course. Like street sign recognition, artwork recognition, I don't know but evolving this might show more potential. Or it might just fail or get overrun by some other gismo millions of hipsters will love and help finance into skynet.

this absolutely. i have nightmares with maps and smaller street signs in japan. there is just so many kanji, and at best i can only read, maybe, 200.

  On 12/17/2010 at 9:31 PM, goDel said:

It's more likely that purely statistical methods (i.e. markov chains) are being used than Chomsky's universal grammar. Chomsky is far too complex to have any real practical value. And apart from it's complexity, it's far from perfect too. Statistical methods often lead to better results in more efficient ways. Chomsky's biggest contribution was putting the science/math into linguistics, not revealing scientific truths in the field of linguistics.

I don't know why you put the word "science" next to "math". It just looks like linguistics are not a genuine science - at least not a logico-mathematic science that can yield universal and timeless truths. And I think that's were is the problem.

When you browse forums about SE technologies, everyone's so hot for the semantic web, but they think semantics are just a matter of keywords extraction or TDIDF analyses, and "IR scientists" replace semanticians. And i tend to think now it's the main tendency in "SE science". They surf on the idea of a world where love is just a chemical equation and the features of a person are defined by her DNA. And ignoring elements brought by linguistics is not going to do any good to the solvability of problems etc...

Also, i think you should have said

  Quote

Statistical methods often lead to better results with less money.

That's why online translators are so bad. They are designed for the masses and thus the algorithms have to be light and fast. But companies like google or wolfram alpha are good at making people believe they are doing high end technology.

And to illustrate what i was previously saying let's try to translate "La pêche a été bonne" with google translate.

"Pêche" is an/a homonym. It is even the archetypical homonym. It both means "peach" and "fishing".

"la pêche a été bonne" is a almost an idiom. Without any supplementary information from the context, it typically means "fishing has been good".

And google outputs the right translation !

What if we add a few words to build up a context so that the sentence now means "The peach was good".

"J'ai cueilli la pêche. J'ai mangé la pêche. La pêche a été bonne."

which i would translate in all humbleness as

"i picked the peach. I ate the peach. The peach was good".

well, this is what google translate returns.

"I picked fishing. I ate fish. Fishing has been good."

And, in regards of signs translation: Google translates "Pêche interdite", as "No fishing" (congrats bro) but "Pêche 3.50€/kg" turns into "Fishing € 3.50 / kg" and "Pêches 3/50€/kg" becomes "Fisheries € 3.50 / kg".

Sort of.

  On 12/19/2010 at 6:43 PM, Babar said:

I don't know why you put the word "science" next to "math". It just looks like linguistics are not a genuine science - at least not a logico-mathematic science that can yield universal and timeless truths. And I think that's were is the problem.

Funny how you immediately use the combination yourself ("are not a genuine science - at least not a logico-mathematic science"). I was merely getting at the fact that Chomsky was trying to describe how a universal grammar would work in terms of your so-called "logico-mathematic" functions. So in that sense, Chomsky was putting the science into linguistics. But it seems to me, you already got that part. I do not agree however that logic or maths are required for something to be a true science, but that's another discussion.

  Quote

Also, i think you should have said

  Quote

Statistical methods often lead to better results with less money.

Sure, if that makes you happy. In real world applications (response)time is an essential factor however, so I'll stick with my use of efficient. Not sure why you would be putting salt on that statement, but whatever floats your boat, i guess.

  Quote

That's why online translators are so bad. They are designed for the masses and thus the algorithms have to be light and fast. But companies like google or wolfram alpha are good at making people believe they are doing high end technology.

"so bad" is a bit harsh, i think. They're not perfect, that's for sure. But what you're trying to prove here? The impossibility of creating "perfect" translators on the basis of either statistical methods or "formal grammar" based methods? It might be impossible, but given the current developments I'd argue it's to early to make such statements.

Back on topic: this app looks like just another glimpse into the future of things which might be possible. For such a cheap application the response-time is quite impressive, and so is the augmented image. It looks like the original, but translated. A couple of years ago I wouldn't have thought these kinds of technologies would be so easily available to the general public and now they are. Sure, it's not perfect, but it looks like developments will go faster and faster.

anyways, whatever.

there was an online translator somewhere that took into account the context and wording of phrases, i forgot the name but it was cool because you could change a comma and it would change the meaning of the phrase.

  On 12/18/2010 at 4:40 AM, remy marathe said:

the tech is cool, but i don't really think this particular application is life changing. speech is a lot more important and i don't think we're anywhere near having a babelfish like device.

No we do actually. There was something that did this even on the previous iphone.

http://www.youtube.com/watch?v=ealQk1lX4yw

Here's what microsoft are up to (vid is from last March actually)

http://www.youtube.com/watch?v=FFe7xVHMl_s

  On 12/17/2010 at 5:10 AM, vproc said:

My hovercraft is full of eels.