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.

Seze

Quote from: Tuiq on March 09, 2010, 11:56:22 AM
To be honest, I don't understand what you're meaning with "part of speech". Isn't (v.), (n.) already added in the English/localized column? It should be quite easy to extract it when you're parsing the file, imho.

Currently the "part of speech" is part of the definition column.  To me it makes more sense to have the "part of speech" as a separate column in the file since its really not part of the definition.  I do agree with you though that it would be easy to parse out.  I just don't see why the "part of speech" is part of the definition when the whole point of the CSV/TSV files in the first place is to have easy access to the raw data for use in other projects.  Its your project, so just tell me what your final choice is on the matter and I will plan accordingly.

Irayo ma Tuiq


Learn Na'vi Mobile App - Now Available

Mithcoriel

Here's some more feedback on the translator:

I misspelled atokirina' , I forgot the apostrophe at the end. Here's the result:

Quoteatokirina (tok) be at, occupy a space Topic. Preposition: Like, as Adjective attributed.

Darn apostrophes, ey. ^^' Problem with Na'vi it's a lot more prone to typos from us sawtute. Maybe it should be even tolerant to missing apostrophes?

Anyway, here's another one that told me to report it: (don't know what "latsu" actually means, I just saw it somewhere)

Quotelatsu (lu) be (predicative copula, existential verb) (am, is, are) Bug. Please report that ats hasn't been found.
Ayoe lu aysamsiyu a plltxe "Ni" !
Aytìhawnu ayli'uyä aswok: "Ni", "Peng", si "Niiiew-wom" !

Tuiq

Quote from: Seze on March 09, 2010, 12:31:30 PM
Quote from: Tuiq on March 09, 2010, 11:56:22 AM
To be honest, I don't understand what you're meaning with "part of speech". Isn't (v.), (n.) already added in the English/localized column? It should be quite easy to extract it when you're parsing the file, imho.

Currently the "part of speech" is part of the definition column.  To me it makes more sense to have the "part of speech" as a separate column in the file since its really not part of the definition.  I do agree with you though that it would be easy to parse out.  I just don't see why the "part of speech" is part of the definition when the whole point of the CSV/TSV files in the first place is to have easy access to the raw data for use in other projects.  Its your project, so just tell me what your final choice is on the matter and I will plan accordingly.

Irayo ma Tuiq

Due to the implementing of the jMemorize-thing I decided to split it apart. It's now Na'vi SEP Native Language SEP Part of Speech NEWLINE.


Quote from: Mithcoriel on March 14, 2010, 05:27:36 PM
Here's some more feedback on the translator:

I misspelled atokirina' , I forgot the apostrophe at the end. Here's the result:

Quoteatokirina (tok) be at, occupy a space Topic. Preposition: Like, as Adjective attributed.

Darn apostrophes, ey. ^^' Problem with Na'vi it's a lot more prone to typos from us sawtute. Maybe it should be even tolerant to missing apostrophes?

Anyway, here's another one that told me to report it: (don't know what "latsu" actually means, I just saw it somewhere)

Quotelatsu (lu) be (predicative copula, existential verb) (am, is, are) Bug. Please report that ats hasn't been found.

We had this whole thing once - ignoring things is bad. Just bad. You maybe know that most browsers have two modes to operate: The standard mode and "oh my god the webdesigner did crap"-mode. Latter is more "I guess you mean that" instead of "You think of this". As a guy who has designed many sites I can tell you, it sucks. Some features are not available if it's in this "guess what" mode and then the whole business gets dirty.

What I want to say: It's not a solution. The actual situation is not much better, BUT (and I like to say that), it gives some interesting views into the language of Na'vi. I'm thinking about a "EXPLAIN THIS WORD" mode that ignores vocabulary (for example, kìyevame would not be 'good bye, see you' but 'k<ìy><ev>ame' which results in see-blah-Icantrememberthatnow).


Oh. And Latsu is fixed. Thanks for reporting.
Eana Eltu: PDF/TSV/jMemorize

Mithcoriel

Well it's up to you. But it might be worth a consideration, if not to make it recognize atokirina as atokirina', to at least make sure it doesn't translate it as some form of "tok".
Ayoe lu aysamsiyu a plltxe "Ni" !
Aytìhawnu ayli'uyä aswok: "Ni", "Peng", si "Niiiew-wom" !

Mithcoriel

Got some more bug reports and similar:

I learned in this post that tsaw = tsa'u . Your translator doesn't know that yet either. Maybe you wanna add it.

Stuff it translated oddly: (from this post: )

satseng --> Unable to translate
I'm guessing Prrton meant the plural of "tsatseng"

oeto (oe) = I, me
[Bug. Please report that AWAY hasn't been found.]

sìomum = Unable to translate
I think Prrton meant the plural of "tìomum", the noun version of "omum".

Another suggestion: How about adding a link to this thread on the page the machine is on? That way it's easier to report bugs and stuff, without searching for the thread anew. (Also, gives you more credit for the machine ;) otherwise people use it forgetting who made it.)
Ayoe lu aysamsiyu a plltxe "Ni" !
Aytìhawnu ayli'uyä aswok: "Ni", "Peng", si "Niiiew-wom" !

Tuiq

"same.equal-THAT" - I think it's a typo, since tsa-tseng is not likely to be cut down to sa-tseng (tsa is '-THAT' - what would sa be then?)

About that away thing - I don't know. It's not gonna be fixed. It may be once the prepositions are loaded automatical. Which can take time.

I don't know if you can just say tì-omum is lenited to sì-omum. There's a lot the translator does not know, and nobody feels like telling me. I don't know anything about the language (anymore), I might not be the best guy to maintain the project.
Eana Eltu: PDF/TSV/jMemorize

Taronyu

Tuiq, for plurals, ay+ is added on the front. However, the + means that it causes the next consonant to lenite - and once this happens, the ay becomes redundant, meaning that a lot of plurals just have lenited initial consonants.

Ts -> s : so tsatseng would go to satseng

Same for tì --> sì.

It's one of the more trouble-some, coding wise, bits of the language.

Tuiq

Well, to be honest - it's not a problem. I just have to change... a few things. The parser has to be modified in a way he supports more-than-one-modifier per postfix/prefix - after that I can tell him that sa- is THAT-LENITED.
Eana Eltu: PDF/TSV/jMemorize

Muzer

This seems to be quite out of date now (unless the link in the first post is old) - are there any newer versions or are you planning on updating this one? Or is it dead now?


(Also, proper nouns don't work which is rather irritating but not too important)
[21:42:56] <@Muzer> Apple products used to be good, if expensive
[21:42:59] <@Muzer> now they are just expensive

Tuiq

Proper nouns are working. They are excluded from the translation system because of the irritating names. And, there's nothing outdated - except the pdf building, because I was thinking the upload to learnnavi.org would work (well, hence, it doesn't.) And since uploading the old files again is a bit stupid, the pdf won't be updated for a few days. I really hope Txepisiyu takes a look soon.

The translation system itself is not really outdated. But I like these comments, "Hey dude, your stuff does not work". It helps really to improve things.
Eana Eltu: PDF/TSV/jMemorize

Muzer

There are quite a few infixes that don't seem to be recognised, as well as a couple of other things like exceptions to rules and
some vowel contractions. It also seems to have problems sometimes with the laudative mood infix (though I've seen it working once...) - eg Oe tsw<er>ay<ei>on (example from Na'vi in a Nutshell).



Also, a bit of more-constructive criticism here (sorry if I've been too vague previously) - you should make it attempt to parse longer words before shorter ones, as a few times (I can't think of any offhand unfortunately) it's picked up on a tiny thing like "lu" in the middle of a big word and told me it's "to be" with loads of endings (including ones usually only suitable for nouns and adjectives). Though you might already do this but due to some of the infixes not working it didn't notice.
[21:42:56] <@Muzer> Apple products used to be good, if expensive
[21:42:59] <@Muzer> now they are just expensive

Tuiq

Exceptions are just evil. Topic closed.

No, seriously. I'll have to take a look at them... later.


About tsw<er>ay<ei>on -

[19:46:04] <Tuiq> !tell tswayon
[19:46:05] <EanaEltu> Tuiq: tsw<1><2><3>ayon

Is the infix position broken? If yes, of course it's wrong. k<1><2>am<3>e for example - kerameie still works.

It has nothing to do with the size of a word. Actually, it is a pretty interesting thing, imho - it shows you what the word COULD be. And while its variation not always makes sens (DAT-GEN-PLURAL etc.) - it /would/ be valid. This is most likely happening with unknown words or typos. There's nothing "buggy" about it - except that I could teach it a few rules (some are already implemented). Believe me, it really tries to use the simplest version possible. If that's not possible it tries to use.. more advanced solutions.
Eana Eltu: PDF/TSV/jMemorize

Muzer

Hmm, must be the few I'd tried were either wrong examples or exceptions... odd...

But yeah, the thing about the infixes still stands - many of the "compound infixes" are unimplemented. I'm mostly using Na'vi in a Nutshell as my reference point here which as far as I understand SHOULD be quite accurate...
[21:42:56] <@Muzer> Apple products used to be good, if expensive
[21:42:59] <@Muzer> now they are just expensive

Tuiq

Lots of infixes aren't included, yes. I was working on a system some time ago but couldn't figure out a nice system - it's still standing in the room after all.
Eana Eltu: PDF/TSV/jMemorize

'eylan na'viyä

Hi Tuiq,

I made a searchplugin for firefox(should work on other browsers too)
so that eana eltu can be used directly from the browser as searchengine

if you add the files, everyone could install the searchplugin directly from the translation interface page.

you only need to copy all files from the .zip(except eana_eltu_ref.html) to the webrootfolder and add the code inside eana_eltu_ref.html to the mainpage.

it includes na'vi -> eng/de/nl/ptbr and an icon i made up for testing (your user image shrinked down to a 16x16 .ico).

Tuiq

Nice. I did everything you said - please check if it's okay.
Eana Eltu: PDF/TSV/jMemorize

'eylan na'viyä

#76
thanks,
everything works fine except the Portuguese version.
it looks like http://dvi.clonk2c.ch/eana_eltu_na_ptbr.xml doesn't exist.
Edit:
seems that the ê in Português caused an xml syntax error

updated version:

Muzer

If you wish to create a search in Opera:

1) Go to the site and select whatever language you want in the list
2) Right click on the main text box
3) "Create search..."
4) Enter a title for the search and a key letter/word if you wish (I chose na, so if I type in the URL bar (for example) na Oel ngati kame it'll translate that.)
5) Click OK
[21:42:56] <@Muzer> Apple products used to be good, if expensive
[21:42:59] <@Muzer> now they are just expensive

Tuiq

Eana Eltu: PDF/TSV/jMemorize

'eylan na'viyä

For Firefox there is a nice addon "add to searchbar" wich works exactly the same as in opera (nice feature btw) but only few have installed it by default. and you can add searchkeys for the locationbar too.
btw: in firefox you can use ctrl+k to select the searchfield and ctrl-up/down to choose the engine for faster access or ctrl+l for the locationbar.