Not to distract from the core of your argument, but "simply tagging words" is still a hard AI problem in general, with all those pesky "time flies like an arrow" vs. "fruit flies like an apple" cases. Of course you are not doing anything innovative in that area if you just call an API provided by someone who did the hard work.
Re:Don't Support Information Architects
by coldtea on Thursday December 26, 2013 @11:23PM
(#6969388)
>Not to distract from the core of your argument, but "simply tagging words" is still a hard AI problem in general
If you do it properly and automatically, yes.
The way Writer Pro (and the underlying Apple API) do it, not so much.
Re:Don't Support Information Architects
by ballard on Thursday December 26, 2013 @10:02PM
(#6969231)
Semantic checking seems far off, but would be an impressive accomplishment:
"Invisible pink bananas sleep furiously."
Don't you mean colorless green ideas?
You're overthinking it.
It's the most naive implementation you can think of. It's exactly this: you select "verbs" and the verbs are highlighted. You select "nouns" and the nouns are. Same for adjectives, adverbs, prepositions (as, from, in, etc) and conjuctions (and, or, etc).
Nothing earth shattering, nothing you can't do with simply tagging words, nothing marchine-learning or AI about it.
Here's an example of Syntax Control in action: http://imgur.com/NrseuZV
Beside the naiveness of the core functionality, notice also how naive the tagging is. As noted below by bosie, "programming" is marked as a verb in the phrase "in programming terms", syntactic context be damned. Same for "run-on" where run is highlighted as a verb too.
P.S Damn, it's even worse than that. All the hard work -- which was not that hard itself -- was already made by Apple in their CoreText API. According to the Verge:
">the heavy lifting is already baked into Apple’s developer platform. Since iOS 5 and OS X 10.7, Apple has provided a class called NSLinguisticTagger that segments natural-language text and labels the text with various bits of information, including parts of speech".
P.S 2 Now, since Apple itself added the capability to tag parts of speech in the iOS API, and since this is a very obvious example of using this (Apple themselves showed something similar in WWDC 2011 IIRC), isn't iA's patent application essentially like asking to restrict the use of Apple's very specific API? It would be interesting to see iA's lawyers fight Apple's lawyers.
P.S 3 This was part of a comment in a sub-thread, but thought it would be more useful posted at the top level to give everybody an idea of the feature.