Mailing List Archive

Spellchecking
Someone knows how fast the aspell/pspell function are? It seems to be easy
(if apache and php are properly configured) to run a spellcheck through a
wikipedia page, which displays misspelled works in a proper color.

Should be used by pressing 'preview', IMHO.

Some Pseudo-Code:

- Create array of words from article text.
- foreach word pspell_check(appropriate_dict, wort)
-TRUE: next
-FALSE: enclose word in color tag

That's without personal spellbooks, replacements and so one, I think
getting wikipedia tags inside the languagedict, or use a additional dict is
enought.

--
------------------------- Anthill inside! ---------------------------
Re: Spellchecking [ In reply to ]
On Sat, May 10, 2003 at 06:55:17PM +0200, Thomas Corell wrote:
> Someone knows how fast the aspell/pspell function are? It seems to be easy
> (if apache and php are properly configured) to run a spellcheck through a
> wikipedia page, which displays misspelled works in a proper color.
>
> Should be used by pressing 'preview', IMHO.
>
> Some Pseudo-Code:
>
> - Create array of words from article text.
> - foreach word pspell_check(appropriate_dict, wort)
> -TRUE: next
> -FALSE: enclose word in color tag
>
> That's without personal spellbooks, replacements and so one, I think
> getting wikipedia tags inside the languagedict, or use a additional dict is
> enought.

Of course it can't possibly work.
Encyclopedia contains too many proper names and other stuff that no
dictionary knows.
Re: Spellchecking [ In reply to ]
Tomasz Wegrzanowski wrote:
> On Sat, May 10, 2003 at 06:55:17PM +0200, Thomas Corell wrote:
>

> Of course it can't possibly work.
> Encyclopedia contains too many proper names and other stuff that no
> dictionary knows.

No way to get a dictionary which will work for everything. It's like the MS
Word(TM) spellchecking: A lot of words marked wrong are spelled correct,
but some easy spelling error can be found. This is of course far from a
perfect spellchecking. It's not supposed to be a perfect one. But it's a
way to get one.

Smurf
--
------------------------- Anthill inside! ---------------------------
Re: Spellchecking [ In reply to ]
On Saturday 10 May 2003 19:38, Thomas Corell wrote:

> No way to get a dictionary which will work for everything. It's like the MS
> Word(TM) spellchecking: A lot of words marked wrong are spelled correct,
> but some easy spelling error can be found. This is of course far from a
> perfect spellchecking. It's not supposed to be a perfect one. But it's a
> way to get one.

IMHO the expected number of false postives will look very bad on the webpage.
On the other hand konqueror the KDE webbrowser now supports in its CVS
version spell checking within the "edit box" making it very easy to spot
typos.

best regards,
Marco


--
Marco Krohn
Theoretical Physics
University of Hannover