Re: [lojban] Re: fosay translator

[Ivo Doko <ivo.doko@gmail.com>]

On 22 April 2011 22:49, Z-CORE <zoltankochan@gmail.com> wrote:
> Sorry, but I did implement it only for declerative sentences yet. And
> it translates
> between English, Hungarian and Ukrainian. Lojban is used in the
> interlingua. E.g.
> the interlingua of "Zoltán Kocsán látott egy házat." will be:
> sentence:
> {
>      clause:
>      {
>           accusative:
>           {
>                noun-phrase:
>                {
>                     quantity: none
>                     difinity: undifinite
>                     noun:
>                     {
>                          order-number: 3
>                          form: formal
>                          real-number: singular
>                          lemma: zdani
>                          noun-type: common
>                     }
>                     persone: third
>                }
>           }
>           verb-phrase:
>           {
>                truth: 1.0
>                verb:
>                {
>                     lemma: viska
>                     order-number: 2
>                }
>                tense: past
>                mood: declarative
>                aspect: indefinite
>           }
>           subject:
>           {
>                noun-phrase:
>                {
>                     quantity: none
>                     difinity: difinite
>                     noun:
>                     {
>                          personal-name: first-name
>                          order-number: 0
>                          form: formal
>                          tags: ukrainian
>                          real-number: singular
>                          lemma: zoltan.
>                          noun-type: proper
>                     }
>                     noun:
>                     {
>                          personal-name: surname
>                          order-number: 1
>                          form: formal
>                          real-number: singular
>                          lemma: kotcan.
>                          noun-type: proper
>                     }
>                     persone: third
>                }
>           }
>      }
> }
> As you can see, Lojban words are used as lemmas. I'm planning to use
> Lojban when
> describing the knowladge base so I will implement the translation for
> it too.
> The translation of the "fuzzy" Lojban phrases to the target languages
> will be a
> set of alternatives. My Lojban is poor yet but I know that there's a
> possibility
> to say a sentence with an undifined tense. Thus, it will give three
> translations
> for such sentences:
> "A girl saw a house."
> "A girl is seeing a house."
> "A girl will see a house."
> Something like that.
>
> On 22 Кві, 23:30, ".alyn.post." <alyn.p...@lodockikumazvati.org>
> wrote:
>> On Mon, Apr 18, 2011 at 01:20:28PM -0700, Z-CORE wrote:
>> > Hello, the Lojban community.
>> > My name is Zoltan Kochan. I'm working on an Interlingual Machine
>> > translator. I've started it for 3 natural languages: English,
>> > Hungarian and Ukrainian.
>>
>> > But I'm a big fan of constructed languages and I've realized that
>> > Lojban can be used as a good pivot-language during machine
>> > translation. That's why I've chosen it's dictionary for representing
>> > meaning during the translation process.
>>
>> > It's not a super-translator... hopefully yet. But it can translate
>> > many types of simple and complex sentences in any! direction between
>> > the added languages. And it has the ability of extending itself with
>> > other (custom) languages. I've "invented" ATNL and CWS for describing
>> > the grammar and the dictionary of any other language.
>>
>> > Anyway, it's an open-source project (https://code.google.com/p/fosay/)
>> > and anybody can download it, try it, have fun with it:)
>>
>> > And I'll be very grateful for any kind of help! Somebody can help me
>> > with filling the dictionaries. Others can support the development
>> > process (It's written in Python 3).
>>
>> Cool!
>>
>> Will you provide some examples of what your program does?  How does
>> it translate the following?  (note that these translations are not
>> precisely identical, I'm not assuming your program flips between them,
>> I'm curious what it looks like to go from Lojban => English and from
>> the English to Lojban.)
>>
>> .i ta mo / what is that?
>> .i ta rokci / that is a rock.
>> .i ta me lo do xu rokci / is that your rock?
>> .i ta me lo mi rokci / that is my rock.
>> .i xu do ponse lo do rokci / do you have your rock?
>> .i mi ponse lo mi rokci / I do have my rock.
>> .i mi djica lo za'i ponse lo do rokci / I want your rock.
>>
>> -Alan
>> --
>> .i ma'a lo bradi ku penmi gi'e du
>
> --
> You received this message because you are subscribed to the Google Groups "lojban" group.
> To post to this group, send email to lojban@googlegroups.com.
> To unsubscribe from this group, send email to lojban+unsubscribe@googlegroups.com.
> For more options, visit this group at http://groups.google.com/group/lojban?hl=en.
>
>

That's a pretty neat idea. Since natural languages are ambiguous the
best thing you can do except for teaching a computer how to recognise
context (which you probably couldn't accomplish successfully in a
reasonable timespan) is to provide several translations, while
"grading" each one with how probable it is.

Is this accomplished through neural network or some other means of
programming artificial intelligence?

-- 
mu'o mi'e .ivan.

-- 
You received this message because you are subscribed to the Google Groups "lojban" group.
To post to this group, send email to lojban@googlegroups.com.
To unsubscribe from this group, send email to lojban+unsubscribe@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/lojban?hl=en.