[lojban] Re: Suggestions Sought: NORALUJV and jbovlaste

[Robin Lee Powell <lojban-out@lojban.org>]

On Thu, Mar 13, 2003 at 09:09:54AM -0500, John Cowan wrote:
> Robin Lee Powell scripsit:
> 
> > > > fazyta'a                annoy+talk
> > > > fanza+tavla:  chatter:   x1 = tavla1 (talk) = fanza 1 (pest), x2 =
> > > > tavla2 (talked to), x3 = tavla3 (talk subject), x4 = tavla4 (talk
> > > > language), x5 = fanza2 (annoyed)
> > OK, we're obviously talking about different standards.
> > 
> > x1 is a chatterer who chatters at x2 about x3 in language x4
> 
> Oh, I understand.  In that case, you want to use the stuff in the parens
> to generate this:
> 
> 	x1 is a talk/pest with talked-to x2 with talk-subject x3
> 		with talk-language x4 with annoyed x5
> 
> This is not totally smooth English, but it is a mere matter of
> copy-editing to make it so.  

True.  But a very large percentage of even the valid lujvo in NORALUJV
don't have even that much information;

 1           ba'ucpi                 utter+bird

Not really fixable, though.

> Break the place structure at the commas, use regular expressions to
> extract the useful stuff, and assemble it according to the template
> "x1 is a ... with ... x2 with ... x3 with ..." etc. etc. It would be
> an easy hack to change the third word to "an" appropriately.

<nod>

> > And, by the way, fanza2 is clearly x2 of fazyta'a (i.e. equal to
> > tavla2), not x5.
> 
> Quite right.  That sort of semantic error *has* to be corrected by a
> human being, of course.

Which gets us back to "which human being, and how?".

-Robin

-- 
http://www.digitalkingdom.org/~rlpowell/    ***    I'm a *male* Robin.
.i le pamoi velru'e zo'u crepu le plibu taxfu
.i le remoi velru'e zo'u mo .i le cimoi velru'e zo'u ba'e prali .uisai
http://www.lojban.org/   ***   to sa'a cu'u lei pibyta'u cridrnoma toi