Re: [lojban] Lojban CFG Questions

[".alyn.post." <alyn.post@lodockikumazvati.org>]

On Fri, Dec 10, 2010 at 10:24:39AM +0100, Roman Naumann wrote:
> Hello dear lojbanistanians,
> 
> i'm starting with an attempt to find a CFG for lojban - or if it shows 
> impossible, to prove it being impossible (so that you see what exactly is 
> impossible and thus to improve).
> 
> Back in 2008 when there was a challenge [1] to do exactly that, i would have 
> done it, but i lacked the knowledge to proof or proof-wrong grammars. Learned 
> it during the last year at university and thought, hey, why not put it to use.
> 
> However, i realized just now looking at the EBFN [2] causes eye cancer. Also, 
> i've never been good with lojban at all. I don't get the (formalizing) problem 
> with elidable terminators yet. To get started, it would be extremely helpful 
> to work on an abstraction. I'd be glad if you could provide such to me. To 
> give you an idea what kind of abstraction i have in mind, here's an example 
> (though perhaps not very useful):
> 
> We have five kinds (sub)sentences. They start and terminate with 'a', 'b', 'c', 
> 'd', 'e'. Inside a (sub)sentence, only subsentences with a letter later in 
> alphabet may stand. ("a c c a" is thus valid, "b a a b" is invalid [whitespace 
> ignored], as a..a is no valid subsentence of b..b).
> Inside of sentences may (beside any number of subsentences) stand zero or more 
> of numbers (which are our abstraction of words). Each number starts with zero 
> and may not contain further zeros (this is to spare us the necessity for 
> whitespace). Terminators [a-e] may be elided, if directly followed by another 
> terminator. Thus, a implicitly terminates [b-e] subsentences, b implicitly 
> terminates [c-e] subsentences and so on.
> A valid example 'word' of the language is: "a 01 02 c 0 e c 08 04 a"
> It should parse to: """ a(01 02 c(0 e())c 08 04)a """
> (didn't want to draw a parse tree, but it this this is enough to get the 
> point)
> 
> So, do you think this abstraction catches the elidable terminator problem or 
> is it too simple? If it's too simple, why, what's missing?
> Besides elidable terminators, are there other problems why you think lojban 
> can't be expressed as a CFG (without the grammar being way too large)?
> 
> Regards,
> Roman
> 
> [1] http://www.mail-archive.com/lojban-beginners@lojban.org/msg04337.html
> 
> [2] http://www.lojban.org/publications/formal-grammars/bnf.300.txt
> 

Roman,

I'm currently working on a parser for Lojban.  I'm not anywhere near
the first person to do so, and I've written my understanding of the
effort on my parser's page:

  http://wiki.call-cc.org/eggref/4/jbogenturfahi

I don't understand enough about the union of your problem and Lojban
grammar to follow your e-mail, I'll need to spend more time reading
it.

If you're able, how about expressing your problem on the "toy"
grammar presented in the e-mail archive from this page:

  http://www.digitalkingdom.org/~rlpowell/hobbies/lojban/grammar/

leads to:

  http://www.digitalkingdom.org/~rlpowell/hobbies/lojban/grammar/jc_mail.txt

which has this example:

  start = sumti selbri
  sumti = LE selbri /KU/
  selbri = tanru | NU tanru /KEI/
  tanru = BRIVLA | tanru BRIVLA

That would help me understand your question better.  It may be that
the tools I'm working on would be of some use to you, if you think
so, I'm happy to discuss here or privately.

-Alan
-- 
.i ko djuno fi le do sevzi

-- 
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.