Hi Matthew,
Thanks for taking time to answer my unqualified observations.
Yes, I am just talking about grammar checkers.
I find it useful, that graviax is
- fast
- simple to use
You are right, if you say, that a lot of grammatic errors
are simply due to one badly used expression or word; to find these
graviax is ideal.
If there were a possibility, that I can say to graviax
using a rule:
Find all expressions using a plural noun after a number,
and mark them as bad, and print the error:
- no plural after a number
(for example: five boys)
or:
Find all expressions, that has an
article- an attribute with plura ending- a plural noun
mark it as bad printing:
- no plural ending of attribute after article and plural noun
(for example: the nices girls)
(The above two are Hungarian grammar rules)
And the like, that is general rules, then graviax could be very well
used for German, Hungarian, and also other languages.
If there are no general rules, just individual rules possible,
I think, graviax is not so useful in spite of its clear
structural advantages.
What do you think about this, is there a chance to extend
graviax to handle more general cases?
Thanks: Eleonora
Thanks for taking the time to have a look at graviax.
> I tested the graviax spell cheker on windows (at present it is only
> for windows).
You seem to be using the terms 'spell checker' and 'grammar checker'
interchangably. Both graviax and Au Gramadoir are grammar checkers.
> My impression is, that it is more a frame for xml applications, than a
> real spell checker. It does not read the words of the language, and it
> makes very limited checking of a few, not general rules.
At present, graviax is more of a 'proof of concept'. You're right
that it is not anywhere near finished -- unfortunately I don't get as
much time to spend on it as I would like.
I've deliberately kept the number of rules low for now, to minimize
the rework if the syntax of the rules needs to change (there are
fewer than 100 rules at present; there could easily be many
thousands). However, even in its current form, I think graviax
highlights some interesting things:
1. It is possible to find a lot of errors using simple pattern
matching rather than part-of-speech tagging. I work as a proofreader,
and want a tool that will allow me to easily create new rules and to
switch sets of rules on and off for different clients. It might be
that graviax should be used as a separate final parse after checking
with a more traditional grammar checker.
2. It is a very good idea to provide descriptions and explanations
with the rules. If a user knows why an error has been flagged up they
can make an informed choice about what to do. If you don't have this,
people don't trust what the tool is telling them.
3. Bundling tests with the rules is a good idea, especially if
advanced users will be encouraged to write their own rules.
4. By using XML and standard regular expressions for the rules files,
they could be reused in other tools and easily read from other
programming languages.
Here is a sample rule:
<!-- before hand -> beforehand -->
<rule match="\bbefore hand\b([^-])"
suggestion="beforehand$1">
<description><dfn>Beforehand</dfn> is one word.</description>
<test in="I would like it before hand, if possible."
out="I would like it beforehand, if possible."/>
<test in="I need to see it before hand."
out="I need to see it beforehand."/>
<test in="I read it before handing it to you."
out="I read it before handing it to you."/>
<test in="It needs testing before hand-over."
out="It needs testing before hand-over."/>
</rule>
[snipped I/O, which just shows that the tool has a long way to go]
> For me this output and therefore the tools capability is far from a
> very useful one at the moment. I would be interested to hear about
> graviax from its author or other testers.
I would also be interested to hear comment from anyone else that has
tried graviax.
> I am looking for a really fast spell checker, that has the capacity of
> Daniel Naber's excellent languagetool. Graviax obviously is not the
> tool of choice at the moment.
>
> I also checked Gramadoir, which for me seems to be extremely hard
> wired to Irish/Gaelic/Celtic and I see no chance to apply that for any
> other language. This is probably due to its lack of usable
> documentation, but also the structure of the program might be the
> reason.
Here are a few situations in which a graviax-like tool would be more
useful than a traditional grammar checker:
1. Converting documents between US and British English. It would be
useful to highlight terms ("first floor", "could care less") and
offer alternatives ("ground floor", "couldn't care less").
2. Checking that references/citations conform to a particular format.
3. Providing sets of rules for particular professions (legal,
medical, ...).
4. Allowing publishers and organizations to standardize their
spellings (e.g. to mandate their choice of 'email' or 'e-mail').
5. To improve users' grammar by teaching them what they are doing
wrong. For example, if they type "I go there everyday" then explain
that it should be "every day" instead, and *why*. That way, they're
less likely to make the same mistake in future.
Best wishes
Matthew
P.S. Has there been any progress on the proposed OOo Grammar API?
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
