[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[lojban] Re: minimal lojban


A minimal Lojban is closely related to what I am doing with the Jorne project. (jorne.org)

1. Start with a simple but expressive subset of Lojban.

2. Reduce each Lojban utterance into a collection of binary (2-sumti) predicates.
(These are called "triples", because they consist of sumti+brivla+sumti.)

3. Convert these triples into a standard format that can be merged with existing data on the web.


I can illustrate the process with the following text:

le prenu cu tavla le skami

This can be changed into binary predicates:
(also see attached graphic)

A isa tavla-instance
A first-sumti B
A second-sumti C

B isa le-instance
B references D

E isa prenu-instance
E first-sumti D

C isa le-instance
C references F

G isa skami-instance
G first-sumti F

At this point, it isn't really Lojban text anymore; but it is still the logical equivalent of the original text, in triples format.
You could use any logic representation for this (such as Prolog).
Jorne uses something called RDF because it is the standard for defining "triples" on the web.

I've got a working RDF equivalent of the above text. It's still experimental, and I currently have to hand parse the text, but it can make queries on a knowledge base and get a list of matching IDs.
Once a parser is integrated, you'd be able to type some minimal Lojban "ma" queries and get a list of IDs for matching items (like sumti, but they could be any information resource).

> ma skami
ID 758346435
ID 753457674

Whenever I can find time, I plan to:
a. Integrate a Lojban parser
b. Return the results as minimal Lojban

Don't want to bore everyone with the technical details, but I'll email sample RDF code to anyone who is interested.

mi'e la iuban.

Robin Lee Powell wrote: 
On Wed, Jan 26, 2005 at 05:27:39AM -0800, Jorge Llamb?as wrote:
  
but i know there are bo and (ke, ke'i)... surely lojban isn't
complete without these??
      
ke and ke'e are very rarely used. 
    
I'm actually starting to use them quite often, oddly enough.

  
bo is more often used in the {i <tag> bo} or the {na'e bo}
constructions than in tanru. In the end it's a matter of taste
whether you think ke-ke'e, bo or co is more essential for lojban.
I wouldn't include bo and ke-ke'e in my minimal lojban, but that's
just my point of view.
    
If a minimal Lojban includes tanru, you *must* have one or the
other, or you lose expressive power.  As it seems to me that the
goas here is to produce a Kernel Language for Lojban (that is, the
most minimal set of Lojban from which all the rest of Lojban can be
described as just shortcuts), you don't want to lose expressive
power.

-Robin



To unsubscribe, send mail to lojban-unsubscribe@onelist.com


Yahoo! Groups Links

JPEG image