Loading...
 
Development

Development

Forum List Topic List

Forums » Development » Thought about spellchecking with Tiki....
 
Thread actions
PDF Print this page
  • Print all pages

    • Thought about spellchecking with Tiki....

    Posted by Xavi (as xavidp - admin) posts: 957 ☆ ☆ ☆

    Hi devs (specially sylvie, but also any other people interested in this issue ):

    I was grading grading users contributions to forums these last days, and I found, as always, plenty of syntactic and grammatic errors on their writing in their mother tonge (such as I may be producing here in English, Catanglish or Spanglish ;-). I was searching again ways to ease their task of spellchecking (at least), and searched around a bit today about this issue. I attach here some comments and links, FYI (this is another issue very important for Tiki use in education, even if it's more important for lower levels of education: primary, secondary, etc... where we make stronger emphasis on correct writing, etc. At uni level, they should write properly... even if most of them don't, yet, un luckily :-/)

    (1) Kspell & Konqueror, KDE, GNU/Linux (client-side)

    First, to inform that under Linux using KDE, there is a nice KSPELL program (using ispell, as far as I understood reading the help rolleyes ) which integrates very well with Konqueror ("The" default browser for KDE environment under Linux, with no equivalent under M$ Windows environment, unluckily). This makes misspelled words appear in red inside the forms where users are writing their contributions (forums, wiki pages, etc.). So far, similar to what Tiki spellchecking feeature produces (when it works, and for the languages it works; I couldn't make it work for a non-english dictionary after making it in the db with the same structure as english dictionary in tiki db cry)

    Besides that, Kspell & Konqueror allow right cliking on any word inside html forms and run spellchecing feature: (add to list, ignore, etc. into local dictionaries), in a similar way as spellcheking feature in OpenOffice.org Writer, or M$ Word, etc.

    See a screenshot here (50kb):
    http://tikiwiki.org/tiki-download_wiki_attachment.php?attId=484

    (2) Missing equivalent feature for Mozilla Firefox (which would be multiplatform: Windows, Linux, etc., client-side)

    Second, I couldn't find a similar engine under Mozilla Firefox so far with latest versions of Firefox (1.0.7 by the time of this post). Which is a shame, because similar integration would make windows users also take profit of this spellchekcing features while working on tiki (or anyother CMS, & html form, of course)

    (3) Tiki-based solutions, beyond Browser based (client side)

    Some wikis have already spellchecking included (such as MoinMoin ), which also add the chance to add misspelled words to some database in the server, to train the spellcheker, etc. I'm not a technician, but looked good enough to me (much better of what is achieved in Tiki righ now, btw). Examples at:


    Moreover, I've seen pre-made php scripts to help spellchecking & replacing strings, such as "spellcheck.php" GPL'd by Chris Snyder. Demo available on-line:
    http://chxo.com/scripts/spellcheck.php
    And source code also there:
    http://chxo.com/scripts/spellcheck.php?showsource=1#source

    Maybe something similar could be added to tiki (once coypright issues are solved? - maybe the auther may like releasing another version of his code under LGPL?)

    (4) Other solutions (server-side)

    Months ago while talking with Sylvie she thought that a call to Aspell (or ispell, I don't remember right now exactly) on the server-side, would help this spellchecking (and also lemmatization, or something similar; sylvie, correct if I'm wrong, please; or even explain what lemmatization means, if different from what aspell & ispell can do already on the server side).

    Some other references

    ASPELL: http://aspell.sourceforge.net/
    Zend Spell checking in Php: http://www.zend.com/zend/spotlight/spellchecking.php#Heading2
    PHP Pspell Functions: http://www.php.net/manual/en/ref.pspell.php

    Then...?

    I'm not a coder. At the most I can do my best to get funding from university projects in which we are using tiki for research on educational scenarios to contribute finnacing the coder/s which want to code it... Anyone intersted? Any possible estimate of expenses (order of magnitude)?

    Cheers,

    Xavi wink

    Reads: 4451
    Link
    Posted by Damian Parker posts: 2881 ☆ ☆ ☆ ☆ ☆ United Kingdom

    > Any possible estimate of expenses (order of magnitude)?
    >

    Nice shiney new Range Rover would cover it I think wink

    Damian

    Link
    Posted by marbux posts: 29 ☆ ☆ ☆

    xavidp, the spellchecker for Firefox is an extension based on MySpell. MySpell ships with the Mozilla browser and Mozilla Thunderbird email client. It is also used in OpenOffice.org, although I'm not sure it is still in the 2.x build that was just released. Anyway, it is available for Firefox under the name Spellbound. I imagine under the standard Mozilla 1.1 license, although it may be LGPL as well, since that is how OOo is licensed.

    On spellcheckers for web apps, I haven't really surveyed this area, although it is on my Todo list. We are going to be integrating several web apps including Tiki and we don't want to inflict more separate spellcheckers on our end users than we have to. So we will be looking for a server side solution. I know there are a few Java-based open source web app spellcheckers, but don't recall most of their names at the moment. One is WBOSS.

    But Aspell probably has broader support than any other open source spellchecker, and certainly has more word lists available. See e.g., Kevin's Word List Page, maintained by the maintainer of Aspell. Aspell also has the advantage of missing fewer errors than other spellcheckers tested. It is the default spellchecker on several Linux distributions, although there is usually a choice between it and Ispell, an earlier Gnu project now replaced by Aspell. There is also a Windows port of Aspell. There is at least one web app spellchecker based on Aspell, Speller Pages. I have not tried it, but it looks from the screenshot that it gives users no option to add words to a custom word list.

    Haven't you been involved in TikiDAV development? I haven't tried it, but I wonder if having your students do their editing in OOo via TikiDAV wouldn't give you the advantage of a solid spellchecker, plus reducing the number of interfaces they need to learn.

    Hope this helps.

    Link
    Posted by Xavi (as xavidp - admin) posts: 957 ☆ ☆ ☆

    Thanks, Marbux.I didn't find the spellchecker for firefox previously. I've just installed from your link. Thanks. However, I thought in a single server side solution would be nice... (mabny of the students' homes, or even computer rooms in educatioanl center have only Windows Internet Explorer installed, for instance... :-/ )

    And yes, TikiDav solves partially one of this problems, if all students use it... (not yet, and TikiDav still bug fixing and evolving to make it more available to all OOo clients, Web servers,etc.). I'm not a coder (Javier Reyes is "the" TikiDav coder, I'm just tester and potential "user").

    Link

    Posted by Rick Sapir / Tiki for Smarties posts: 3665 ☆ ☆ ☆ United States

    How about a quick fix so that the spellchecker doesn't see capitalization as a difference. Currently on my 1.9 TIKI, although the word foo may be in the database, if a user types Foo, the spell checker flags it as a mispelling.

    Short of duplicating the entire spelling DB table, is there a quick fix for this?

    -Rick

    Link
    Posted by marbux posts: 29 ☆ ☆ ☆

    > How about a quick fix so that the spellchecker doesn't see capitalization as a difference. Currently on my 1.9 TIKI, although the word foo may be in the database, if a user types Foo, the spell checker flags it as a mispelling.
    >
    > Short of duplicating the entire spelling DB table, is there a quick fix for this?


    While I don't know the specifics of the Tiki spellchecker, I am familiar with the technology. What you are describing is a feature not a bug. I can't imagine why anyone would want a case-insensitive spell-checker. Don't you want to be able to catch capitalization errors?

    If it is only an occasional word that bothers you, just add the capitalized word to your word list from the spellcker dialog. Boom, it won't bother you again.

    But if you are bent on creating a case-insensitive spellchecker, you will need to edit the word list and duplicate in the file all of the capitalized words as lower case and vice versa. Any heavy duty text editor will have case conversion capabilities. Any words you later add to the word list will have to receive similar treatment.

    But I would back up that original word list file before you start. I doubt if you will like the result.

    Link
    Posted by Rick Sapir / Tiki for Smarties posts: 3665 ☆ ☆ ☆ United States

    > While I don't know the specifics of the Tiki spellchecker, I am familiar with the technology. What you are describing is a feature not a bug. I can't imagine why anyone would want a case-insensitive spell-checker. Don't you want to be able to catch capitalization errors?
    >

    A capitzlization error is not a spelling error. Why should the word first be flagged as a misspelling, just because I started a sentence with First.

    > If it is only an occasional word that bothers you, just add the capitalized word to your word list from the spellcker dialog. Boom, it won't bother you again.


    Yes, but there is no easy way (that I'm aware of) to duplicate the entire word list. I, as an end-user, am forced to entere each "new" word into the word list each time — this is very unuser-friendly. mad


    >
    > But if you are bent on creating a case-insensitive spellchecker, you will need to edit the word list and duplicate in the file all of the capitalized words as lower case and vice versa. Any heavy duty text editor will have case conversion capabilities. Any words you later add to the word list will have to receive similar treatment.
    >
    > But I would back up that original word list file before you start. I doubt if you will like the result.

    Link