Re: Spelling Checker

bryan vold (bvold@ingenix.com)
Fri, 30 Jul 1999 16:37:27 -0500


I forgot to mention a couple of things. The spell checker is written in
Java (but could obviously be adapted to other computer languages). Also
for the curious, the other corrections offered by the phonetic portion of
the spell checker for the word laff are: "laugh", "aleph", "leaf", "aloof",
"loaf". This may give you some insight as to the algorithm that it uses.
I've also experimented with Ternary Trees, Patricia Trees, DoubleArray
implementations of Tries, and Dual-Double Array implementations (haven't
had the time to code these last two yet but am excited about the speed-up
potential /efficiency of these structures).

-Bryan

On Fri, Jul 30, 1999 at 03:11:20PM -0500, bryan vold wrote:
> I also wrote a phonetics-based spell checker. Essentially a word is broken
> up into "sound pieces" and the pieces are then replaced by the
> super-classes of sounds it could represent, and then try all known
> (defined) spellings of these sound pieces on these words. I am very
> interested in this area and would be interested what approach you took to
> creating a suggested word ranking. One has to be a little careful around
> this as IBM owns a patent on this technology, which they've tried to make
> as all-encompassing as possible. If you are interested, perhaps we could
> trade some ideas and/or code. I'm thinking about releasing what I've done
> so far to the world under an as-yet-undetermined license. Let me know what
> you think.
>
> -Bryan
>
> PS
> My checker will correctly return "laugh" as a candidate for laff.
>
> On Wed, Jul 21, 1999 at 12:20:24PM -0700, Josiah Carlson wrote:
> > I have noticed in the most recent built that is downloadable from the
> > abiworld web page, that there is a spelling checker, but not a suggestor.
> >
> > For a CS class that I took this past spring, I wrote a spelling
> > checker/suggestor that I believe could be put to good use. Namely in the
> > fact that it is fast and that it has good suggestions that other programs do
> > not come up with (*point and laugh at Microsoft Word*), even with a smaller
> > dictionary of words.
> >
> > If someone could possibly acquire a plaintext version of a word dictionary
> > that is as large as others (MS word has a 64k word dictionary), I would be
> > willing to donate my source (I need to modify it a tad, for it was
> > originally written to use a poorly constructed word class (damn integration
> > classes).
> >
> > I would also be willing to include custom external header files so that the
> > integration between the sources could start as soon as I get them made up,
> > rather than having to wait until I finish with the source.
> >
> > If you are interested, please respond as soon as you get the chance.
> > Thanks.
> >
> > Josiah Carlson
> >
> >
> >
> >
> > _______________________________________________________
> > Get your free, private email at http://mail.excite.com/
>
> --
> bvold@ingenix.com

-- 
bvold@ingenix.com


This archive was generated by hypermail 1.03b2.