> A much-hacked IBM5170. In other words a PC/AT.
Admittedly I have the
> later version motherboard with the 16MHz clock crystal. There is a
> 486-kludgeboard in place of the 80286 processor so the thing can run
> linux.
On Sun, 24 Nov 2013, Jules Richardson wrote:
Well, you just need a second one, which can then
handle all of the
spell-checking load :-)
The OED is available on CD-ROM!
Version 1 seems promising. It's just the full text, and could presumably
be accessed with anything that you can rig up that can read the CD-ROm
file system.
Unfortunately, version 2 apparently added ridiculous copy protection that
never worked right, and made it unusable.
SOME OEDs are surprisingly reasonable on eBay!
I saw a version 1.1 of the CD-ROm for about $100! BUT, does 1.1 have
that f'ing copy protection disaster?
There were also very nice deals on the OED compact edition.
(NOTE: do NOT confuse that with the "concise" edition, which is a marginal
dictionary. The "compact" edition is the real thing, but photographically
reduced to put 4 or 9 pages on each side of each sheet)
Are there any spell-chequers available without auto-correct, but which
could flag words not in their lexicon (reverse video?)
Context sensitivity might also be useful, although even that won't always
do the job. On the closed captions on the TV, I just saw a mention that
"due to static, astronauts would have to be lexically grounded". Do ANY
spell-shequers provide lexical grounding??!? I could really use THAT.
One of the obvious problems with spell-chequing for folk such as us, is
that our vocabulary is not "standard".
30+ years ago, I designed, but never finished, a heuristic system. The
idea was to start out with a VERY tiny word list, or even none at all.
Every word that it didn't recognize would require Correct, Ignore (leave
it, but it's not something I'll use again), or Add to word list.
It would NOT have been successful in the market! Besides the hardware at
the time being inadequate in both speed and access of the list, the
general public would have HATED it.
1) People judge a program based on their first couple of sessions with it.
And that system would be miserable for the first few sessions, with
interaction required for damn near every word. Over time, the word list
would grow until it was an almost exact match for YOUR writing. But until
then, it would be slower and more work than commercial systems that only
include "ordinary" words.
2) It assumes that you KNOW the correct spelling, and simply need a
reminder when you hit wrong keys. It is NOT suitable for people who don't
have a clue how to spell anything, and expect MICROS~1 WEIRD to tell them.
3) A word list comparison, without context sensitivity is useless for
people who don't even know what the correct word is, and don't know the
difference between "accept" V "except", "too" V
"two" V "to", "their" V
"there" V "they're", etc.
That also plays into the issue of why a GIANT word list doesn't solve the
problem. If "hte" and/or "teh" were words in the wordlist, then the
spell-chequer would be useless.
d) Does not make sill corrections without informing
me.
Although "sill" was probably not his intent, it IS prob'ly gonna
be in the
word list.
4) specialized vocabulary.
Tony would want a lot of technical terms.
and things such as signal names.
Some people would want some colloquialisms. Should "ain't" be in the word
list? I remember trying out Manfield's "Electric Webster, and being told
that "deli" is "NOT A VALID WORD".
Should it enforce a preference for "till" V "until" V
"'til" V "til"?
Should the word list include "MicroSloth"?
Why SHOULDN'T the word list include proper names that you use regularly?
I also tried to adapt similar heuristic algorithms to OCR - hire the
neighbor's kid to type everything in red on the graphics display; after a
few pages, MOST OCRs successfully; after a lot of pages, you've
successfully added any consistent font (NOT handwriting!)
I sure had a lot of projects that I never finished, and never will. :-(