Re: Editing AppleWorks Dictionary?

[bird@MCS.COM (J. L. Walters)]

In article <69.48529432.Logic.IS@logicbbs.org>,  <Jim.Low@logicbbs.org> wrote:
>One thing I haven't tried is just elimating the standard AppleWorks main 
>dictionary, and using an exclusive custome dictionary.  Of course, it would 
>mean creating a very large custom dictionary.

Three problems with that approach:

1. The dictionary would be limited to 16,250 lines. The supplied
dictionary has 90,000 words.

2. The verify speed would go way down since the custome dictionary has
to be searched lineraly (sp?) which means N**M/2 (24,000,000 for a
3,000 word document) compares in the worst case. Putting your most
common words towards the top would help.

3. The effort to generate such a beast, making sure that all added
words were spelled correctly.

The problem (as I remember the explanation from a friend who has
disassembled AppleWorks), is that there are just a few "words" in the
dictionary e.g., A, is, the, etc., etc. The rest are "formed" by
AppleWorks matching a prefix and then finding a matching suffix e.g.,
prefix: "com" would match with e and mb.

As I remember the explanation it is much more complex than above since
one prefix can lead to a concatenation of the prefix with a subsequent
one and this new prefix is then matched with a whole new set of
endings and if this fails the added prefix is stripped off and the
hunt continues.

I (wrongly?) view the process as one of decending a tree where the
prefix is made up of the names of the folders as we decend and the
successful match takes place when the word being checked matches the
concatenation of the pathname and one of the files that the pathname
points to.

So, in order to do what you want would require:

1. A program(e) ;-) that would break out all of the possible words
   into an ASCII file.

2. Manually go through and removes all the words that are spelled
   wrong in the culture under discussion and add those that are
   missing. 

3. Another program would then have to repack the whole mess back
   up. (Can you say, "BIG awk/perl/icon program boys and girls?")

Kind Regards,


-- 
	Joe Walters:  bird@mcs.com or J.L.Walters@genie.com
	The sound of your horn is far sweeter if blown by somebody else.