Eana Eltu: Translator, Dictionary, API and putxìng.

Started by Tuiq, January 07, 2010, 04:20:17 PM

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.

Tuiq

Interested in translating the dictionary to your language? Check this out

EanaEltu is the project to generate Taronyu's dictionary, allow people to translate it to their native language (or even Na'vi), the API that allows developers to access this data easily and generate apps that are always up to date or provide data as TSV/jMemorize/(My)SQL (notice the license, cc by-sa-nc). As an example application there were a translation and a dictionary/wordlookup long time ago as webapps as well as an IRC bot some of you may know.

Note: The source code of the majority of EE is available at github. This includes used example plugins to generate output (to add new file formats, as example). The SQL file is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 3.0 Unported License.
Eana Eltu: PDF/TSV/jMemorize

Ftiafpi

!!!!! WOW This is amazing, I put in: Tsun oe ngahu pivängko a fi'u oeru prrte' lu and it translated it perfectly! Wonderful!

Irayo, irayo ma tsmukan, fi'u nìngay srayung!

Hawnuyu atìtse'a

Somebody sticky this. One of the more useful tools I've seen, and easy to understand. Irayo!
"And that's how you scatter the roaches."- Col. Miles Quaritch.

Team Quaritch Member

Rey ulte ting rey.

omängum fra'uti

It seems to get tripped up on infixes not in the basic vocabulary such as <iv> and <us>.  I tried throwing in the simple "Oeru txoa livu" and it couldn't translate "livu".  Other ways I tried to trip it up it seemed to handle, as long as it was sensicle to begin with.  (ex. it didn't know what to do with le-tì-fpom, but that's just silly to begin with.)
Ftxey lu nga tokx ftxey lu nga tirea? Lu oe tìkeftxo.
Listen to my Na'vi Lessons podcast!

Ftiafpi

Yeah, doesn't like infixes but that's okay, just take them out and translate those words in your head.

omängum fra'uti

It handles other infixes fine, just not <iv> and <us>.  As for me, I'd just translate myself to begin with, it's probably quicker half the time anyway unless I want to keep yet another browser tab/window always open.  I'm more thinking in the terms of people who are still a bit new and not used to spotting infixes/root words.
Ftxey lu nga tokx ftxey lu nga tirea? Lu oe tìkeftxo.
Listen to my Na'vi Lessons podcast!

Tuiq

Quote from: omängum fra'uti on January 07, 2010, 04:59:31 PM
It seems to get tripped up on infixes not in the basic vocabulary such as <iv> and <us>.  I tried throwing in the simple "Oeru txoa livu" and it couldn't translate "livu".

If you can give me an explanation what <iv> or <us> do (or what they are called) or a link to a reference I'll implement it. I'm not really sure about the infix positions, at the moment it has only two, for example:
t<>ar<>on. Like I said, at the moment it's really based on the pocket guide.

Quote from: omängum fra'uti on January 07, 2010, 04:59:31 PM
Other ways I tried to trip it up it seemed to handle, as long as it was sensicle to begin with.  (ex. it didn't know what to do with le-tì-fpom, but that's just silly to begin with.)

Actually I didn't know if it's possible to have that kind of construct - I assumed "no" so I didn't teach him that. However, it should work now.
Eana Eltu: PDF/TSV/jMemorize

Night Raider

Is this right? I just write whatever in the signature box?

Payoang

Incredible work... just, wow. Can I link to it from the main site? I'd love to host the program on LearnNavi.org, too, of course.

omängum fra'uti

Quote from: Tuiq on January 07, 2010, 05:14:20 PM
Actually I didn't know if it's possible to have that kind of construct - I assumed "no" so I didn't teach him that. However, it should work now.
Well as I said it's somewhat nonsensical, since it's already a noun to begin with so the "tì" is redundant, then making a noun THEN an adjective seems just downright silly rather than just going straight to adjective.  I was just seeing what it did with it.

As far as <iv> and <us>, the best information we have right now is that they are the subjunctive and participle respectively.  There is also an <ev> infix but we don't yet know what it does.  It seems like it would be at least doable for it to record the infixes we know exist but don't know or don't positively know the meaning of, even if it's just with a comment about it not being fully understood.
Ftxey lu nga tokx ftxey lu nga tirea? Lu oe tìkeftxo.
Listen to my Na'vi Lessons podcast!

Tuiq

Quote from: omängum fra'uti on January 07, 2010, 05:23:44 PM
Well as I said it's somewhat nonsensical, since it's already a noun to begin with so the "tì" is redundant, then making a noun THEN an adjective seems just downright silly rather than just going straight to adjective.  I was just seeing what it did with it.
Yes, but this would also affect if you try to make a verb-to-noun-to-adjective - however, it can't hurt having le-ti as a prefix.

Quote from: omängum fra'uti on January 07, 2010, 05:23:44 PM
As far as <iv> and <us>, the best information we have right now is that they are the subjunctive and participle respectively.  There is also an <ev> infix but we don't yet know what it does.  It seems like it would be at least doable for it to record the infixes we know exist but don't know or don't positively know the meaning of, even if it's just with a comment about it not being fully understood.

I've added them with a comment that they are not really known yet. As I said, I may have to rewrite a bit if it's possible to have things like t<us><im>ar<ei>on since actually it only knows t<>ar<>on.

Quote from: Seabass on January 07, 2010, 05:20:36 PM
Incredible work... just, wow. Can I link to it from the main site? I'd love to host the program on LearnNavi.org, too, of course.
Of course you can link it. However, I want to clean up the code a bit first and improve it - there's still a lot of performance to gain.
Eana Eltu: PDF/TSV/jMemorize

Swok Txon

Terrific job! that would be been hard work!

irayo tsmukan!

txum tukru

awesome work, can there be a vice versa option, like, English to navi :)
pesu nga?          "who are you?"
Oe lu toktor.       "the Doctor!"
pesu?                "who?"
nì'aw, toktor.       "just, the Doctor!"

Tuiq

I've posted it in another thread already - most likely not. Since Na'vi is a beautiful constructed language (there are no known, non-explanation-able things like irregular verbs, times or something like that, the word order doesn't matter, accusative etc. is defined as a postfix) it is really easy to parse it - whether English is not. As an example from the IRC Channel:
Quote
Dangerous Norman doesn't like to be a blue brain

A Na'vi translation would be "Lehrrapa Norman längu eana eltu" - but where do you know that? Sure, there's S-V-O in English, but even then, lookup 'dangerous' is not possible since it's not defined (of course - danger is defined, you could try to lookup 'danger'+ous, 'danger'+y etc - but it's not the same as in Na'vi).

However, if you would enter something like "danger-ADJR Norman be.TIME(PRESENT)<PEJ> brain blue.ADJL" it's easy parse- and doable. Since having this kind of "primitive" form in your head when translating Na'vi anyway this is really easy. It's on my todo.
Eana Eltu: PDF/TSV/jMemorize

Taronyu

Let me know if you want the newest versions of the dictionary, or if I can make any edits to any particular codes that would make it easier for you to compile from (or whatever the term is).

Tuiq

Quote from: Taronyu on January 08, 2010, 10:33:00 AM
Let me know if you want the newest versions of the dictionary, or if I can make any edits to any particular codes that would make it easier for you to compile from (or whatever the term is).

I would always be glad about the newest version. Your code is really fine, the LaTeX is easy to parse. A thing that I've just seen and that's a bit annoying is that not everything seems to have a gender - for example the first thing, 'ampi - touch. Noun or verb? If it's a verb, it isn't recognized as one yet and thus cannot be conjugated with infixes. There are a few words in the dictionary that have this problem. It's not that bad, but at least every verb should be identified as one. If you could add 'v.' as a gender it would be very useful.

In order to get a properly working English->Na'vi translation the problem will not really be your dictionary - it will be to parse the English language. However, if I find something that could be helpful I'll let you know. I'm starting with it tonight, as soon as I've finished a few improvements that are flying around right now in my head.

Also: Since it seems like you know the grammar quite a bit (I've read your Inflections PDF), it says that <us> is in the first position (t<1><2>ar<3>on?) - what about <ev>?
Eana Eltu: PDF/TSV/jMemorize

Taronyu

<iv> goes in the second position. Only <us>, as far as we know, takes the first. About gender, more appropriately called parts of speech - where it has been removed is where we don't know whether it is a noun or a verb. And part of speech I've given is mere guesswork, as the original document did not specify.

I'll upload the newest version somewhere, if you like.

Tuiq

Quote from: Taronyu on January 08, 2010, 12:01:32 PM
I'll upload the newest version somewhere, if you like.

Yes please. Or send it by mail - I think you should still have my address.
Eana Eltu: PDF/TSV/jMemorize

Tuiq

Update: Now using 0.028 of Taronyu's dictionary. Added a few things, fixed some.
Eana Eltu: PDF/TSV/jMemorize

Wayutral

Awesome! A few days ago I was considering coding a similar app, but I don't need to worry about it anymore ;)