Parser/Glosser

[WorldMaker <world_maker@xxxxx.xxxx]

I have an idea for working out a quick and dirty Keyword Glosser that
could theoretically get the point across. You would need two parts (both
of which could be externalized) a multi-dimensional lookup table and
"Parsing Rules". Since there are multiple keywords that will work for a
word. Let's take menli to start with:

menli [ men ] mind ; 'mental' 
x1 is a mind/intellect/psyche/mentality/[consciousness] of body x2 

You would have the first most used keyword as the Zeroth item in the array
(and the default) and intellect as number 1 and so on.

Then, if you wanted to refer to intellect instead of mind you write 
menli[1] instead of just plain menli.

Its a rough way of doing it, but theoretically, if you can get some sort
of standard going on the order of the keywords and then you can do lookup
tables for other languages.

Last, you would need "Parsing Rules" to figure out how the sentence would
appear. For instance it could look something like the example sentences
at the bottom of the definitions and you could probably generalize it to 
"%1 is a(n) %k [of body %2]" for english and which the parser would
substitute the first sumti for %1 and the keyword (from the lookup table)
fot %k and if the second sumti exists it includes the part in brackets and
substitutes the second sumti for %2.

You externalize this into say "ENGLISH.TBL" and then all you have to do is
rewrite the table as a different file ("RUSSIAN.TBL") and rules for
another language.

What do you think? Its just an idea I had, and there are probably a bunch
of odd problems that would creep up and you would have to write parser
rules for different sumti configurations (%1 would still refer to the
first sumti in order from left to right, but if the parser finds a
modifier it uses a different parser rule).

--[mu'ezba]--
{{That works doesn't it?}}
_________________________________________________________
Do You Yahoo!?
Get your free @yahoo.com address at http://mail.yahoo.com