I would be interested in such a comstruction for Dutch, just towarn about
incorrect word splitups.
Ruud.
That's it, thank you!
However, what we have now in Language Tool is a long set of rules, one
for each pair of words. I thought about a tool that would allow adding
new pairs of words to
Laszlo once told me suggestion generation takes a lot of time, and is
programmatically limited in time.
Several options in the .aff limit option and time too.
Ruud
W dniu 2012-04-29 02:12, Daniel Naber pisze:
On Samstag, 28. April 2012, Marcin MiÅkowski wrote:
My impression was that it's
Stemming and analyzing, especially getting a postag for a compond not in
the dic, would be great.
Impact is quite large though; for Dutch, tuning of postags to make the
hunspell and LT ones compatible requires revision of some tags.
Good plan by the way...
Ruud
Hi again,
I've just split the
Might be usefull rules for Dutch as well
But there is no Dutch contributor on Java level.
Hi,
Hi, I have done the translations.
I was surprised by the new text warning for identical words and adverbs
at the start of the sentence.
Seems to me that 'same word' overlaps 'same adverb'.
Daniel,
I succeeded in creating prototype rules from the annotations file, a bit
like your easy rules editor.
600 rules in one run. None has been tested yet though. That is the next step.
Ruud
--
Live Security Virtual
The way Hunspell reacts to quotes and dashes is specified in the affix file.
For Dutch, Marcin once adjusted the tokenizer to accept certain quotes. In
other languages, this will result in two tokens from LT, which are by
themselves incorrect.
The dash is controlled in Hunspell as a default
Marcin, in reaction to your quote below, I can only agree that the
computational part is quite complex.
There was some research in Tilburg, resulting in a quite simple
computational algorithm:
- reduce all characters to their simple form (drop accents)
- get their ascii value, raise it to the
There is a words list for Dutch at www.opentaal.org. Relatively small,
just 450.000 words, including proper names, but it is the best open one
there is.
To be exact:
http://opentaal.org/bestanden/doc_download/18-woordenlijst-v-210g-bronbestanden-
Ruud
Ruud
Marcin MiÅkowski
What is the error the text 'Duplicate rule file!' stands for?
What is exactly going on when this message applies?
There is only one grammar.xml, or not?
When not, what makes 2 files duplicate?
Ruud
I'm going to create the LanguageTool 1.8 release tomorrow, so today is the
last chance to
I Like 1, because it links to the implementation perception of users
Maybe combined with 2, which the advantage of already shoming an
applicable icon
Having an underline under languagetool suggests that the name is wrong.
That is not intended of course.
7 shows the multi-lingual aspect quite
That is strange, that should not be the case.
Could you send me the dump?
Ruud
On 14.10.2012, 21:44:59 Ruud Baars wrote:
Though the input for fsa surely has 13-jarig as a base word for
13-jarige
Well, it doesn't. If I export the dutch.dict with this command:
./dump.sh
You could dump it. Daniel did that for me lately.
Think about the postagging, done by the routines in LT which supports
compounds.
Ruud
Hi,
I noticed today that the german.dict file in LanguageTool is a binary
file, I suppose created with Morplogik. Is the original data and the
conversion
Could you send me the text string please, so I can translate them?
Ruud
Is there anyone out here good enough to add this rule to Dutch LT?
We might move some rules from the xml to this rule, maintaining
functionality, and expand it easier and faster.
I've just added
Daniel,
Maybe I am wrong,but suppose this speeds things up, wouldn't it also make
things a lot more complex?
Is it a real option?
Wouldn't a faster server, faster compiler (if available) be easier solutions?
Or could moving some processing to the client side be an option?
Ruud
On
Could you add the link to the translation location?
The extension has been preliminarily reviewed[1] by Mozilla, and is
now available at addons.mozilla.org (AMO)[2]. Please note that this
version is not up to date, so if you want to test the extension, use the
attached version.
I plan to
I have translated Dutch as good as possible.
Ruud.
The extension has been preliminarily reviewed[1] by Mozilla, and is
now available at addons.mozilla.org (AMO)[2]. Please note that this
version is not up to date, so if you want to test the extension, use the
attached version.
I plan to
Fine for me. Makes a big rewording for existing rules necessary though.
Ruud
2013/1/16 Dominique Pellé dominique.pe...@gmail.com
Do we really need to put suggestion inside suggestions?
It would be less noisy like this:
messageyada yada yada/message
suggestionxxx/suggestion
That is redundant functionality imho, prone to introduce differences in
replys.
It is acceptable for migration purposes nonetheless.
Ruud
2013/1/17 R.J. Baars r.j.ba...@xs4all.nl
Fine for me. Makes a big rewording for existing rules necessary though.
Ruud
No need. It will be just another
I guess this is language independ, so be fixed in same rule as no space
before comma.
Ruud
All,
I could fix this bug by coding the following rule within the Italian
grammar.xml (I also added a check for semicolon
I know, I could have used
a reg_exp :-) ).
The question is, would this be
there are more chars this might apply to
:;,.)}
The other way around: there should be a space before some other chars.
Ruud
--
Minimize network downtime and maximize team effectiveness.
Reduce network management and
About case:
In Dutch, DVD is a undesirable way to write dvd; Dvd at the start of a
sentence. Camelwords are proper names most of the time.
Nieuw-Zeeland is one word, and correct. As is auto-deur.
As far as I am concerned, for any word, only a first-uppered form is
acceptable as a variant of any
the firstUpper is a feature for spell
checking, the other for grammar, as long as there is still a distinction.
But why does the grammar not signal the ' start of sentence' to the spell
checker?
Ruud
2013/4/8 R.J. Baars r.j.ba...@xs4all.nl
About case:
In Dutch, DVD is a undesirable way to write
I don't understand most of the technicalities, but I can stat that for
Dutch, multiple single and multiple character substitutions are necessary
to get the proper word.
Example: cocoskoekie is very informal an wrong for kokoskoekje ( twice the
common c-k interchange, once the common kie-kje
I really do think it is not a good idea to expand a dic using an aff,
unless it is just a munch compressed dic.
Using compouning flags to generate all kinds of theorètical words will
also result in unlikely or wrong words to appear, and will make the list
unnecessary long.
Ruud
Marcin, Hunspell is (almost) perfectly sound, assuming there is a
reasonable text as input, and all features are used extensively.
The issue is that compounding itself is not safe; it is a language issue.
Doing 'expansion' , the process is the other way around: all words, even
words never
API-compatible too?
Hi,
maybe we could also have a command-line interface compatible with
ispell/aspell/hunspell? This way, we would have a drop-in replacement
for these spellers.
Best,
Marcin
W dniu 2013-05-05 17:33, Daniel Naber pisze:
Hi,
I created a page that describes how to use
The handling of dots is mostly done in the calling app, and quite
differently between apps.
But when spelling woudl be called as part of one language support module,
not 2 separate spelling and grammar module, it would be possible to handle
this correctly for both.
Ruud
What's the best
Some words have tat much ways of being used, that almost every postag is
valid.
These words that require a of of exceptions, or, which is better,
pre-treatment in the disambguation.
Ruud
Hi,
I have created my first rule. It somehow works, but its behaviour is
weird. The rule should find a
So there is no construction like referring to an a ?
Anyway, there are wasy to work around it.
Ruud
DÅa 25.06.2013 11:14, R.J. Baars wrote / napÃsal(a):
Some words have tat much ways of being used, that almost every postag is
valid.
These words that require a of of exceptions, or, which
Disambiguation will help.
Use words before and after the letter to rule noun out.
Ruud
On Tue, Jun 25, 2013 at 12:35 PM, Milos Sramek
sramek.mi...@gmail.comwrote:
DÅa 25.06.2013 11:14, R.J. Baars wrote / napÃsal(a):
Some words have tat much ways of being used, that almost every postag
Coding word frequencies as a character is fine. I think it would be
classes, logarithmic as far as I am concerned.
Ruud
W dniu 2013-07-16 00:03, Jaume Ortolà i Font pisze:
2013/7/15 Marcin MiÅkowski list-addr...@wp.pl:
Hi Jaume,
W dniu 2013-07-15 21:16, Jaume Ortolà i Font pisze:
Hi,
When you are going to need postags, I would be happy to co-operate on
Turkish; I can provide a Turkish words list and word frequencies.
Ruud (Dutch)
Hi everyone,
My name is Sinan Yolal. I do not know so much about Java and I am learning
XML. If it is okay, I would like to add Turkish to
You are right.
It is right there for download.
Ruud
I think there was a complete dictionary and a tagger in the abandoned
project we mentioned earlier... But I may be wrong.
Best,
Marcin
W dniu 2013-07-19 08:09, R.J. Baars pisze:
When you are going to need postags, I would be happy to co
This is really very useful!
Jaume
I am all for moving useful pieces into common code, please implement
the necessary changes - I'll be on vacation starting tomorrow without
permanent Internet connection so I may not get to do that myself for
next couple of weeks.
Thanks
Andriy
Daniel, for non-developers like Github is indeed a blocking thing.
Developer tools are just too complicated.
A web interface to contribute to rules is the best may to go for xml
rules, I guess. But then, there has to be some form of levels of access, a
workflow from developing a rule to test,
I have reported the low improvement rate for Dutch to the OpenTaal
community, which planned to take over actions from myself.
if that does not lead to a repsponse soon, I will take over the LT Dutch
again myself.
I am quite busy nowadays, but the least I can do is remove the false
alarms, which
You could use the postag SENT_START on the first token
Ruud
I thought this would ignore occurrences of actually|really in the
beginning of a sentence:
pattern
token/
token regexp=yesactually|really/token
token/
/pattern
But it doesn't. How do I
Yes, it is a snapshot of a few days ago, with edited disambiguation.
Tests run perfectly.
I can send the changed disambiguation, maybe it is in that.
But tonight, I am in the train right now and then at work, without
connections. Security...
Ruud
On 2013-09-29 22:32, Ruud Baars wrote:
No, that is all I changed, but with a (recent) snapshot.
I will check tonight with the new release.
- without the altered disambig
- with it
It could be platform, But I am not the only one on (K)ubuntu, using JDK.
Ruud
On 2013-09-30 18:17, Ruud Baars wrote:
For reproducing exactly: remove
No need for excuses. imho this is already quite stable. I will download a
snapshot and start again.
I am glad to have helped getting rid of a bug.
Ruud
The BaseSynthesizer is now thread-safe. The PolishSynthesizer had the
same bug and has been fixed, too.
@Ruud:
Sorry for the
It might be possible to generate Hungarian postags from the info in the
Hungarian Hunspell.
Ruud
On 2013-10-09 10:40, Logon wrote:
Hi Karoly,
I am new to LanguageTool, have just learned about it as the HU
translator of
OmegaT.
I wonder if it would be possible to start specifying rules
I just noticed the long process of having lT digging through the Dutch
input files has succesfully ended without any further crashes.
Looks like the fixes applied were great!
Ruud
--
October Webinars: Code for
Does anyone know the maker of the Afrikaans spell checker?
It seems to have a lot of proper names without the required capital.
Ruud
--
October Webinars: Code for Performance
Free Intel webinars can help you accelerate
languagetool-graphical ?
languagetool-user-interface?
Ruud
Hi,
a user was confused about the name languagetool-standalone.jar[1].
Does anybody see a problem with renaming the file to
languagetool-desktop.jar? languagetool-gui.jar might even be more
appropriate, but I assume not everybody
Thanks for adding this check.
I am currently just too busy, but will trty to find the time to correct
these rules.
Ruud
Dominique Pellé dominique.pe...@gmail.com wrote:
Hi
I've just added new checks when running ant check. It now
reports the following new errors in Russian, Chinese and
Sounds like regexp is applicable? Or are there just too much?
Hello!
There are some special verbs in Portuguese which affect the verb that
comes next, for example: ter/haver.
Variations of the verb ter to add to the rules:
1) tinha
2) tinhas
3) tinham
4) tenha
5) tenhas
6) tenham
7)
As discussed before, there is actually no real connection between language
and country.
These codes could (and should) not be handled as one setting at all.
English is one of the languges spoken worldwide, with small variants for
regions like Australia, New Zealand, Great Britain, USA etc.
Ruud
It is huge (10 GB). Since it is data captured from sites, it is not
guaranteed to be as free as is needed for free publication of results.
The format is plain text, utf8, 1 line per original paragraph.
Ruud
On 2013-11-26 11:49, R.J. Baars wrote:
This is a big help if it also works on other
The corpus data does not have to be free, as long as you don't reproduce
the data, do your own process and add intelligence.
Anyway, I collected rather reliable frequency data from Internet for lots
of languages. I am willing to 'free' the counting data on some reasonable
conditions.
These
Why show sentences with errors to people that need help getting it right?
It is not an objection, more a question of : is there a reason from the
user perspective?
Ruud
Hi,
we have example sentences for every error in grammar.xml, and we show
them on community.lt.org, but not in the
There are 2 mechanisms in Hunspell, the one using compound rules and the
one using compound-start , -middle and -end.
Dutch uses both, rules for the easy system of numbers etc, the other one
for regular compounds.
One cannot mix those two methods.
Other languages don't compound at all, most
Marcin,
Is there a readable document that shows how prefixes and suffixes are
defined and applied in the morfologik speller?
Ruud
Hi all,
I want to introduce compounding support for MorfologikSpeller for the
next LT release. I looked at hunspell dictionaries again and it seems
that it
My perception is that Hunspell suffers from the short flag notation.
Longer flags would allow for more flags, thus reducing the need for
conditions for applying the affixes (the condition is pre-applied by
applying (longer) flags to the right base forms.
Maybe complexity in the mechanism could
Marcin,
I assume most of the entries in the dictionary will be words, acceptable
without any affix or other additions.
In that case, a flag 'notaword' could be used to keep non-word parts to be
accepted as words. A bit like FORBIDDENWORD in Hunspell.
Ruud
I read the part about configuring the speller dictionary of morfologik.
Some aspects seem strange to me. Looks like some or all options allowed
are for the entire list only (?)
Case-specificness is important. Now, I notice some hunspell dictionaries
accept compounds that are incorrect imho. I
Weird.
On 2014-01-08 15:26, R.J. Baars wrote:
Case-specificness is important. Now, I notice some hunspell
dictionaries
accept compounds that are incorrect imho. I noticed German accepting
BurgerInnen.
That's a feature I think. BürgerInnen is sometimes used as a short
form of BÃ
Could someone please this text from the wiki?
Interactive testing of rules using a corpus
My account does not allow edit.
Ruud
--
CenturyLink Cloud: The Leader in Enterprise Cloud Services.
Learn Why More Businesses
Inflection can be done, using the 'inverse dictionary'.
If there is a Lithouanian postag dictionary, in generating this from the
data, the reversed dictiuonary can be made, and the 'inflected' can be
used.
There are examples in the wiki as well as some languages, like Dutch.
Ruud
Hello to all
I guess this depends on the case-specificity of the dictionary entries.
Are they, or aren't they?
hendrix should be in the dictionary as Hendrix, not hendrix, so the
suggestion should always be uppercased, it being a proper name.
DVD is a very common error (in Dutch) for dvd, so DVD should
In fact, I suggest only lowercasefull stopUppercase should be reported.
Ruud
Daniel Naber wrote thus at 06:18 PM 19-02-14:
This rule will cause new false
alarms for words like X.Org.
If it's only when the first character after the dot is in cap, then it's
okay.
Otherwise, we've a lot to
But Marcin is right. There should not be a noun after this expression, but
most likely an adjective.
Ruud
I understand that U/C part, but more or less in
this case is an idiom meaning to a varying or
undetermined extent or degree (Merriam-Webster's
Collegiate® Dictionary). We can't have it
I would be glad to update translations, but there is only the English text
to start from, no context of the text at all, or is there?
The same text maybe should be translated differently for different contexts.
Ruud
Hi,
Please update your translations for LanguageToolFx at transifex[1]
The .txt is in the LT distribution, isn't it? I tis generated when the
dictionary was tagged.
(Has not been done since ages).
If it is there, I am willing to add explanations to use until the new tags
are available.
Ruud
Hi,
in order to link POS tag set descriptions from the online rule
I downloaded audio and video. The video is in a bit strange resolution for
Video : 1120*832
This could already cause som problems.
Should I place it in the middel of of a full HD fram, or squeeze it in dvd
format?
KDENLIVE is quite good at this, generating all sort of open formats and
keeping
The .avi seems to be compressed using a quite lossy codec, so the screen
is rather fluffy. It does not seem to be of the same size the one Jan
used.
Is the raw file still available?
Ruud
--
Is your legacy SCM system
21:38, R.J. Baars wrote:
Is the raw file still available?
Now at http://danielnaber.de/tmp/lt-rule-editor-screencast-no-sound.ogv
Regards
Daniel
--
Is your legacy SCM system holding you back? Join Perforce May 7
Would it be possible to:
- add a level of 'checking profiles'
- that have the familiar categories
- the categories just contain rule(set) id's (file_name+rule_id)
- the rule-id's are stored in a number of rule files
Of all profiles, just 1 can be active for the active language. This makes
it
Hi.
I am currently using languagetool-commandline checking billions of
paragraphs.
It works fine, except for some dumps, like the ones below.
I need the tool to continue, for I need the data. When it has been
processed, I might try to find the items it crashes on.
It looks like it is all
Unfortunately, the dump is still there in the nightly build of yesterday:
I will try to find examples making it dump, though they are rare.
Ruud
Exception in thread main java.lang.RuntimeException:
java.util.concurrent.ExecutionException:
java.lang.StringIndexOutOfBoundsException: String index
By the way, it would be helpfull if in the dump there was some info of the
input ... I dont knwo if that is possible, but it would surely help.
Ruud
W dniu 2014-05-27 12:42, Daniel Naber pisze:
On 2014-05-27 11:06, Marcin MiÅkowski wrote:
maybe it was because of a simple mistake in the
It has been running for some hours now; no dump yet.
Ruud
On 2014-05-28 07:28, R.J. Baars wrote:
Unfortunately, the dump is still there in the nightly build of
yesterday:
Are you sure you're using the latest build? The line numbers look like
Marcin's fix is not included. I have added
I did Dutch as far as the core is concerned. I don't know the other
features, So I had to leave those untranslated.
Hi,
this is reminder that we're now in feature freeze for LT 2.6, to be
released 2014-06-30.
* Please help complete the translations at
There is loads of them in the wiki, in the other languages files etc.
It is quite simple, instead of checking the word itself, you check the
'type of word'.
Ruud
But, how?
Is there any documentation around with examples?
On 08/07/2014 07:15, R.J. Baars wrote:
I think the best way
Graag gedaan.
m.v.g.
Ruud
Isn't this the kind of thing
On 2014-07-25 17:03, Elanjelian Venugopal wrote:
I am trying to capture an error pattern where a letter is attached as
a suffix to a POS tagged word. And the suggestion should strip of the
suffix.
I haven't tested it, but something
There is an adjustment to make in the sentence splitter. But where did the
.srx go?
I detected an abbreviation that is commonly used and as for now seen as
sentence end:
milj.
Could this be added to the Dutch srx rules?
Ruud
Same applies to [0-9]{1,2}[-]pers.
Ruud
There is an adjustment to make in the sentence splitter. But where did the
.srx go?
I detected an abbreviation that is commonly used and as for now seen as
sentence end:
milj.
Could this be added to the Dutch srx rules?
Ruud
On 2014-08-18 16:16, R.J. Baars wrote:
There is an adjustment to make in the sentence splitter. But where did
the
.srx go?
It's at
languagetool-core/src/main/resources/org/languagetool/resource/segment.srx
Could this be added to the Dutch srx rules?
Sure, could you send a patch?
Regards
I think a good addition for LT would be to have a general rule, just
acting on tokens, a bit like srx does wit letters.
bed : ok
bed english : not ok = bad english
A mechanism, that lets the longer token list overrule the shorter one.
This would create the option to add found errors empirically.
I am currently checking the output of all the rules on the 20GB corpus;
some rules are perfect, some less (though hard to tweak).
Result will be a major update, I guess...
Ruud
On 2014-08-18 17:18, R.J. Baars wrote:
I was able to test, and removed 2 of my additions to make it work.
Thanks
This generates an error. How do I add multiple urls to 1 rule?
urlhttps://onzetaal.nl/taaladvies/advies/instandhouden-in-stand-houden/url
urlhttp://taaladvies.net/taal/advies/vraag/412/in_bedrijf_stelling_inbedrijfstelling//url
Ruud
This is a limitation; there are several texts to refer to sometimes ...
Is it on the whishlist to change this?
On 2014-08-19 12:41, R.J. Baars wrote:
This generates an error. How do I add multiple urls to 1 rule?
Only one URL per rule is currently supported.
Regards
Daniel
:14, R.J. Baars wrote:
bed : ok
bed english : not ok = bad english
For some types of errors, I think it works better then current
rule/exception type of check.
I'm not sure I understand: do you suggest a different (more compact) way
to write down simple rules, or do you suggest
for Dutch.
Ruud
On 2014-08-19 13:43, R.J. Baars wrote:
hand some; wrong; handsome
I hand some tools to; correct
It is a very compact way of defining very simple rules.
I see, but the thing is that these rules probably won't stay simple for
long. What if you want to add you hand some
Persian or Farsi?
Hi,
some people will already have noticed it: LT has just added support for
Persian. This is an important step, as Persian is the first
right-to-left language we support. There will probably be some bugs, but
for now it's looking good and there have only been minor issues.
java -jar languagetool.jar
(process:14246): GLib-CRITICAL **: g_slice_set_config: assertion
'sys_page_size == 0' failed
This occurs after clicking on the reference url in the UI; it works though.
Ruud
--
Okay, if you need sentences to test on, or plain words lists with
frequencies, or word groups and their frequencies, I have them.
Ruud
https://en.wikipedia.org/wiki/Persian_language
On Tue, Aug 19, 2014 at 5:33 PM, R.J. Baars r.j.ba...@xs4all.nl wrote:
Persian or Farsi?
Hi,
some
I discovered that the rule below is not working very well.
It look like 'skip' also skips over sentence boundaries.
Is that intentional? Or is something else wrong?
In case it is intentional, is there an option to forbid that?
Ruud
rule id=nr738 name=duur kost
pattern
token skip=4duur/token
THanks, I will check it out.
The rule is not functioning very well at all. I commented it out and put
it on the list of items to do.
Ruud
R.J. Baars r.j.ba...@xs4all.nl wrote:
I discovered that the rule below is not working very well.
It look like 'skip' also skips over sentence boundaries
Is it possible to get the entire input in the output? Same for the matched
part?
Ruud
--
Slashdot TV.
Video for Nerds. Stuff that matters.
http://tv.slashdot.org/
___
I would appreciate if it could be added. It makes it a lot easier to cycle
testsets.
I guess I should make a request for it?
Ruud
On 2014-08-20 09:47, R.J. Baars wrote:
Is it possible to get the entire input in the output? Same for the
matched
part?
I think that's not possible without
, but the
context can be very long.
Using the truncated context as new input triggers new errors and misses some)
Ruud
On 2014-08-20 10:15, R.J. Baars wrote:
I would appreciate if it could be added. It makes it a lot easier to
cycle
testsets.
I guess I should make a request for it?
What exactly would
This one will generate false errors on the plain type F16 and mp3.
Not that general after all...
Ruud
I mean these rules are useful for all languages for example
---
should have space between word and numbers numbers
(\d+)(\w+)
$1 $2
will do this action
foo123 foo 123
I checked
I will try to build it using the web interface.
Ruud
On 2014-08-20 10:15, R.J. Baars wrote:
I would appreciate if it could be added. It makes it a lot easier to
cycle
testsets.
I guess I should make a request for it?
What exactly would you need as output? Adding another output format
I try to warn for longer words in uppercase (considered bad habit in
Dutch), but this rule hits far too much.
However, this generates an enormous amount of positives, so I want to make
some exceptions.
One exception should be
UPPERCASE - words words words (newspaper start of line
The other one
I want to catch a double negation (not . not), but want to forbid the
match when there is a certain token in between (like ;).
Is that possible? How?
Ruud
--
Slashdot TV.
Video for Nerds. Stuff that matters.
The texts on unification on the wiki look pretty difficult.
Is this the way to check:
'leger des heils' (salvation army) and correct it into
'Leger des Heils'?
There are quite a lot of proper names that need this kind of correction,
preferrably in one rule ..
What are the options?
Ruud
I have found a solution, but it is a bit of a workaround:
I changed the rule to just check 'leger des heils' case insensitive, so it
will always hit, no mater the case.
To prevent alarm for the correct 'Leger des Heils' I added a case
sensitive rule to the disambiguator.
Ruud
The texts on
Could anyone please elaborate on the layout of:
wrongwordincontext?, Maybe using bad/bed as an example?
I am not sure which of the fields is what.
The german example is not clear to me (Not very good in German).
word1=bad
word2=bed
match1=?
match2=?
context1=?
context2=?
explanation1
explanation2
I stopped using the disambiguator, and used antipattern instead. This way,
the rule is completely in one file.
Thanks for tipping.
Ruud
I have found a solution, but it is a bit of a workaround:
I changed the rule to just check 'leger des heils' case insensitive, so it
will always hit, no
Thanks I added it to the Dutch datafile as well. And one case to start
with. Taht works fine.
But it does not trigger when none of the context words is present. Is that
by intent?
Ruud
On 2014-08-22 07:33, R.J. Baars wrote:
Could anyone please elaborate on the layout of:
wrongwordincontext
1 - 100 of 273 matches
Mail list logo
