Re: prototype of new rule editor
W dniu 2014-04-05 12:04, Marcin Miłkowski pisze:
W dniu 2014-04-04 19:43, Daniel Naber pisze:
On 2014-04-04 10:32, Marcin Miłkowski wrote:
Another one: adding a second example (under Chrome) creates incorrect
XML:
Can you specify the exact workflow that leads to this? I couldn't
reproduce so far.
I don't remember it anymore :(
I'll see if it happens again.
OK, it happened again:
!-- Polish rule, 2014-04-05 --
rule id=ID name=
pattern
token regexp='yes'.*\p{Ll}+ie/token
token regexp='yes' spacebefore='no'['`’]/token
token regexp='yes'e?(?:go|mu|m)/token
/pattern
messageTen wyraz odmienia się prawdopodobnie tak: suggestionmatch
no=1/match no=3 regexp_match=.*(go|mu|m)
regexp_replace=$1//suggestion./message
urlhttp://so.pwn.pl/lista.php?co=Chaplin/url
shortBłąd ortograficzny/short
example type='incorrect'Zaprośmy markerCharlie'go/marker. Zaprośmy
Charlie'ego./example
example type='correct'Zaprośmy Charliego./example
example type='incorrect'Zaprośmy Charlie'ego./example
/rule
What I did was to edit manually two examples, change the rule, and then
add another example, change the rule, and it was there. I'm sure I only
changed the suggestions (match parameters to be exact).
Regards,
Marcin
--
___
Languagetool-devel mailing list
Languagetool-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/languagetool-devel
Re: prototype of new rule editor
W dniu 2014-04-03 21:51, Daniel Naber pisze: On 2014-04-03 21:27, Marcin Miłkowski wrote: The URL box is overly sensitive to the format of URLs, and does not accept for example this one: http://poradnia.pwn.pl/lista.php?id=9687 Thanks, will be fixed with the next deployment (it was a Chrome-only issue). Another one: adding a second example (under Chrome) creates incorrect XML: example type='incorrect'Obiecano odmrożenie markerzablokowanych aktyw/marker reżimu. Nie tkniesz moich aktyw!/example example type='correct'Obiecano odmrożenie zablokowanych aktywów reżimu./example example type='incorrect'Nie tkniesz moich aktyw!/example Note that Nie tkniesz moich aktyw was added as a second example. Then it is concatenated to the 1st example, and there's no marker in the example either. Regards, Marcin -- ___ Languagetool-devel mailing list Languagetool-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/languagetool-devel
Re: prototype of new rule editor
W dniu 2014-04-04 10:32, Marcin Miłkowski pisze:
W dniu 2014-04-03 21:51, Daniel Naber pisze:
On 2014-04-03 21:27, Marcin Miłkowski wrote:
The URL box is overly sensitive to the format of URLs, and does not
accept for example this one:
http://poradnia.pwn.pl/lista.php?id=9687
Thanks, will be fixed with the next deployment (it was a Chrome-only
issue).
Another one: adding a second example (under Chrome) creates incorrect XML:
example type='incorrect'Obiecano odmrożenie markerzablokowanych
aktyw/marker reżimu. Nie tkniesz moich aktyw!/example
example type='correct'Obiecano odmrożenie zablokowanych aktywów
reżimu./example
example type='incorrect'Nie tkniesz moich aktyw!/example
Note that Nie tkniesz moich aktyw was added as a second example. Then
it is concatenated to the 1st example, and there's no marker in the
example either.
And when parsing this XML:
!-- Polish rule, 2014-04-04 --
rule id=ID name=dd
pattern case_sensitive='yes'
token inflected='yes' regexp='yes'\p{Lu}\p{Ll}+/token
token-/token
token inflected='yes'zdrójexception
inflected='yes'Zdrój/exception/token
/pattern
messageW dwuczłonowych nazwach miast oba człony piszemy wielką
literą: suggestionmatch no=1/-match no=3
case_conversion=startupper//suggestion/message
urlhttps://pl.wikipedia.org/wiki/Pomoc:Powszechne_błędy_językowe/url
shortBłąd pisowni wielką i małą literą/short
example type='incorrect'Mieszkam
markerRabce-zdroju/marker./example
example type='correct'Mieszkam w Rabce-Zdroju./example
/rule
I get reports that the rule is problematic. It is not.
Also, updating the examples is buggy: when I changed examples to create
a second rule (a variant of the first one), the old example stayed in
the XML.
Regards,
Marcin
--
___
Languagetool-devel mailing list
Languagetool-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/languagetool-devel
Re: prototype of new rule editor
On 2014-04-04 10:32, Marcin Miłkowski wrote: Another one: adding a second example (under Chrome) creates incorrect XML: Can you specify the exact workflow that leads to this? I couldn't reproduce so far. Regards Daniel -- ___ Languagetool-devel mailing list Languagetool-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/languagetool-devel
Re: prototype of new rule editor
W dniu 2014-04-02 16:27, Daniel Naber pisze:
On 2014-03-17 17:51, Daniel Naber wrote:
there's now a prototype of a new rule editor available at
http://community.languagetool.org/ruleEditor2/. Main features are:
I have released another update. Major new features:
-Parse existing XML link to get an existing XML rule into the editor.
This doesn't support everything, but at least it should tell you which
element is not supported in those cases.
-Attributes of tokens and exceptions can now be set, even if the editor
doesn't know about them ('skip' is an example)
-Small user interface improvements
The URL box is overly sensitive to the format of URLs, and does not
accept for example this one:
http://poradnia.pwn.pl/lista.php?id=9687
which is of course correct.
Regards,
Marcin
--
___
Languagetool-devel mailing list
Languagetool-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/languagetool-devel
Re: prototype of new rule editor
On 2014-04-03 21:27, Marcin Miłkowski wrote: The URL box is overly sensitive to the format of URLs, and does not accept for example this one: http://poradnia.pwn.pl/lista.php?id=9687 Thanks, will be fixed with the next deployment (it was a Chrome-only issue). Regards Daniel -- ___ Languagetool-devel mailing list Languagetool-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/languagetool-devel
Re: prototype of new rule editor
On 2014-03-17 17:51, Daniel Naber wrote:
there's now a prototype of a new rule editor available at
http://community.languagetool.org/ruleEditor2/. Main features are:
I have released another update. Major new features:
-Parse existing XML link to get an existing XML rule into the editor.
This doesn't support everything, but at least it should tell you which
element is not supported in those cases.
-Attributes of tokens and exceptions can now be set, even if the editor
doesn't know about them ('skip' is an example)
-Small user interface improvements
Regards
Daniel
--
___
Languagetool-devel mailing list
Languagetool-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/languagetool-devel
Re: prototype of new rule editor
On 2014-04-01 16:06, Marcin Miłkowski wrote: Heh, even worse, one cannot specify suggestions, only messages. See the comment below the 'Message' box: Use 'single quotes' to mark words that will be shown as suggestions to the user. Adding suggestion elements outside the message isn't supported indeed. Regards Daniel -- ___ Languagetool-devel mailing list Languagetool-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/languagetool-devel
Re: prototype of new rule editor
W dniu 2014-03-24 20:02, Daniel Naber pisze: On 2014-03-17 17:51, Daniel Naber wrote: there's now a prototype of a new rule editor available at http://community.languagetool.org/ruleEditor2/. I've deployed another update: -more help texts, but they can be hidden by advanced users -fixed the jumping cursor bug that occurred in Chrome -improved way to display the analysis of the example sentence in case of a problem (i.e. no error found in wrong example sentence) There's one feature I need: I frequently need to take an older rule and test it. Just because there is Lucene index online, the online editor is nicer, so I used the old interface for this (in the advanced mode). But the new editor does not give me anything like that, and I actually would like to have more: the ability to paste the rule and see it rendered in the dialog. I know this is not the critical feature but it's definitely helpful for debugging rules that have become complex over time. Regards, Marcin -- Learn Graph Databases - Download FREE O'Reilly Book Graph Databases is the definitive new guide to graph databases and their applications. Written by three acclaimed leaders in the field, this first edition is now available. Download your free book today! http://p.sf.net/sfu/13534_NeoTech ___ Languagetool-devel mailing list Languagetool-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/languagetool-devel
Re: prototype of new rule editor
On 2014-03-17 17:51, Daniel Naber wrote: there's now a prototype of a new rule editor available at http://community.languagetool.org/ruleEditor2/. I've deployed another update: -more help texts, but they can be hidden by advanced users -fixed the jumping cursor bug that occurred in Chrome -improved way to display the analysis of the example sentence in case of a problem (i.e. no error found in wrong example sentence) Regards Daniel -- Learn Graph Databases - Download FREE O'Reilly Book Graph Databases is the definitive new guide to graph databases and their applications. Written by three acclaimed leaders in the field, this first edition is now available. Download your free book today! http://p.sf.net/sfu/13534_NeoTech ___ Languagetool-devel mailing list Languagetool-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/languagetool-devel
Re: prototype of new rule editor
On 2014-03-23 01:19, Kumara Bhikkhu wrote: Minor issue: Is the A example sentence in the 2 text boxes deliberately wrong? Yes, although I've now changed it to an example that's hopefully a bit clearer. Regards Daniel -- Learn Graph Databases - Download FREE O'Reilly Book Graph Databases is the definitive new guide to graph databases and their applications. Written by three acclaimed leaders in the field, this first edition is now available. Download your free book today! http://p.sf.net/sfu/13534_NeoTech ___ Languagetool-devel mailing list Languagetool-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/languagetool-devel
Re: prototype of new rule editor
I love it! Amazing work. Minor issue: Is the A example sentence in the 2 text boxes deliberately wrong? kb Daniel Naber wrote thus at 12:51 AM 18-03-14: Hi, there's now a prototype of a new rule editor available at http://community.languagetool.org/ruleEditor2/. Main features are: * Checks the example sentence against known errors so nobody wastes time writing a rule that already exists * Has text analysis (POS tags, lemmas, chunks) integrated * Checks rule against a part of the Wikipedia/Tatoeba corpus to help avoid false alarms The basic workflow idea is to start with two example sentences, a wrong one and its corrected version. A (trivial) pattern is then generated automatically, which is just the word(s) that differ in the wrong and corrected sentence. The user then needs to add more tokens to make the rule complete. Finally, it is checked against Wikipedia/Tatoeba. Several things are not supported yes, but please give it a try anyway. Regards Daniel -- Learn Graph Databases - Download FREE O'Reilly Book Graph Databases is the definitive new guide to graph databases and their applications. Written by three acclaimed leaders in the field, this first edition is now available. Download your free book today! http://p.sf.net/sfu/13534_NeoTech ___ Languagetool-devel mailing list Languagetool-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/languagetool-devel
Re: prototype of new rule editor
On 2014-03-17 17:51, Daniel Naber wrote: there's now a prototype of a new rule editor available at http://community.languagetool.org/ruleEditor2/. Main features are: Another new version has been deployed with small changes: * the languages' tagset.txt is now linked, explaining the part-of-speech tags * each match can now be modified (regexp_match and regexp_replace in XML) Please try using this for your rule development, even if you don't need it because you know the XML syntax. Only by actually using it we can find out which of the missing features are the most important ones and where the remaining bugs and usability problems are. Regards Daniel -- Learn Graph Databases - Download FREE O'Reilly Book Graph Databases is the definitive new guide to graph databases and their applications. Written by three acclaimed leaders in the field, this first edition is now available. Download your free book today! http://p.sf.net/sfu/13534_NeoTech ___ Languagetool-devel mailing list Languagetool-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/languagetool-devel
Re: prototype of new rule editor
On 2014-03-17 22:20, Marcin Miłkowski wrote: The old editor does not support unification, and neither this one. This is really bad for languages that use this tool (btw, German could really have easy disambiguation using unification; I can help). What would adding support for unification mean? Just adding unify, similar to marker and setting 1 to n features with optional types? Or also editing the unification elements? Regards Daniel -- Learn Graph Databases - Download FREE O'Reilly Book Graph Databases is the definitive new guide to graph databases and their applications. Written by three acclaimed leaders in the field, this first edition is now available. Download your free book today! http://p.sf.net/sfu/13534_NeoTech ___ Languagetool-devel mailing list Languagetool-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/languagetool-devel
Re: prototype of new rule editor
W dniu 2014-03-18 14:31, Daniel Naber pisze: On 2014-03-17 22:20, Marcin Miłkowski wrote: The old editor does not support unification, and neither this one. This is really bad for languages that use this tool (btw, German could really have easy disambiguation using unification; I can help). What would adding support for unification mean? Just adding unify, similar to marker and setting 1 to n features with optional types? Or also editing the unification elements? Just adding unify. Most unification elements remain unchanged, unless bugs are found. I don't remember right now but I think the same limitation (no unification) is in languagetool-wikipedia SEARCH feature. Regards, Marcin Regards Daniel -- Learn Graph Databases - Download FREE O'Reilly Book Graph Databases is the definitive new guide to graph databases and their applications. Written by three acclaimed leaders in the field, this first edition is now available. Download your free book today! http://p.sf.net/sfu/13534_NeoTech ___ Languagetool-devel mailing list Languagetool-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/languagetool-devel -- Learn Graph Databases - Download FREE O'Reilly Book Graph Databases is the definitive new guide to graph databases and their applications. Written by three acclaimed leaders in the field, this first edition is now available. Download your free book today! http://p.sf.net/sfu/13534_NeoTech ___ Languagetool-devel mailing list Languagetool-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/languagetool-devel
Re: prototype of new rule editor
Okay, that would do it. Verzonden van smartphone. Daniel Naber daniel.na...@languagetool.orgschreef: On 2014-03-18 14:00, R.J. Baars wrote: tokenon/token token postag='AJn|VBp' postag_regexp='yes'/token Checking the rule against 100 sentences is a good check for false positives (though I would like a bigger sample, say 1000). That's not possible for performance reasons. But: those 100 sentences are selected in a clever way. For example, for your rule we'll only check sentences that have the word on and a POS tag AJn or VBp. Regards Daniel -- Learn Graph Databases - Download FREE O'Reilly Book Graph Databases is the definitive new guide to graph databases and their applications. Written by three acclaimed leaders in the field, this first edition is now available. Download your free book today! http://p.sf.net/sfu/13534_NeoTech ___ Languagetool-devel mailing list Languagetool-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/languagetool-devel -- Learn Graph Databases - Download FREE O'Reilly Book Graph Databases is the definitive new guide to graph databases and their applications. Written by three acclaimed leaders in the field, this first edition is now available. Download your free book today! http://p.sf.net/sfu/13534_NeoTech ___ Languagetool-devel mailing list Languagetool-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/languagetool-devel
Re: prototype of new rule editor
On 2014-03-17 17:51, Daniel Naber wrote: there's now a prototype of a new rule editor available at http://community.languagetool.org/ruleEditor2/. Main features are: A new version has been deployed. Major changes: * more example sentences can be added * URL and short message (both optional) can be set * help when writing the error message: typing \1 etc will open a new selection where one can select the case conversion. 'Single quotes' will be turned into suggestion elements. Regards Daniel -- Learn Graph Databases - Download FREE O'Reilly Book Graph Databases is the definitive new guide to graph databases and their applications. Written by three acclaimed leaders in the field, this first edition is now available. Download your free book today! http://p.sf.net/sfu/13534_NeoTech ___ Languagetool-devel mailing list Languagetool-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/languagetool-devel
Re: prototype of new rule editor
W dniu 2014-03-17 20:44, Daniel Naber pisze: On 2014-03-17 19:15, gulp21 wrote: Some suggestions that came to my mind while testing the editor: * add a link to the documentation of the POS tags of the chosen language (or even better: offer autocompletion) Or even better: stop treating POS tags as strings, so we don't need regex anymore for them. For example, a plural noun could be noun number=plural. Developing such a structure isn't trivial, though. I think the only thing we need to think through is how to represent this cleanly in XML. The rest is rather simple, and requires just parsing structural XML. The old editor does not support unification, and neither this one. This is really bad for languages that use this tool (btw, German could really have easy disambiguation using unification; I can help). * warn when a word looks like a regexp or POS tag and the wrong type is chosen Done for regex (not yet deployed). I haven't mentioned yet that we're using a new technology for this editor: AngularJS. While the back-end is still developed in Grails (thus, it's basically Java), Angular makes developing the front-end much easier. We even have unit tests for the front-end now! If you have ever used Javascript, I strongly recommend you have a look at http://angularjs.org. Especially when you hate Javascript. Huh, I can't believe it. ;) Regards, Marcin Regards Daniel -- Learn Graph Databases - Download FREE O'Reilly Book Graph Databases is the definitive new guide to graph databases and their applications. Written by three acclaimed leaders in the field, this first edition is now available. Download your free book today! http://p.sf.net/sfu/13534_NeoTech ___ Languagetool-devel mailing list Languagetool-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/languagetool-devel -- Learn Graph Databases - Download FREE O'Reilly Book Graph Databases is the definitive new guide to graph databases and their applications. Written by three acclaimed leaders in the field, this first edition is now available. Download your free book today! http://p.sf.net/sfu/13534_NeoTech ___ Languagetool-devel mailing list Languagetool-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/languagetool-devel
