Re: Modules for individual supported languages?
W dniu 2013-10-03 21:24, Jan Schreiber pisze: Somebody by the name of Łukasz Janik posted this to our Facebook wall: prosze kazdy jezyk jako osobno I don't speak a single word of Polish, but according to Google Translator, this is a feature request to release single-language versions of LT. (Google and I might be wrong here of course.) ;-) You're right. Łukasz Janik has been pushing this for years now ;) I tend to agree with him. Given the fact that the vast majority of people probably doesn't actively use more than three languages, we're imposing a huge overhead on our users. I'm not so sure that the overhead is so huge, given that the broadband user base is growing every year. We've discussed this before, but I'm not sure what the outcome was. I think the ideal solution would be if the users could configure the languages they want before downloading. If that is not possible, there should be a clean way to remove unwanted languages during or after installation. Maybe we could have a two-step download: In the first step, you download the main app, perhaps with English already on board. During install, you can choose whatever other languages you may need. The easiest way for (Libre|Open)Office users would be to have separate downloads for every language, just like with spelling dictionaries. This is already feasible due to modularization. We can already create this via Maven poms. Also, some languages could even offer spelling hooks to (L|O)Office to replace the deadly slow hunspell. Same goes for the Firefox extension, and the standalone app. The only remaining problem is that for bilingual rules, we really need some mechanisms to communicate between the modules, and to download modules on the fly. Office users don't use that, but for CheckMate (translation QA) that could be a problem. Regards, Marcin -- October Webinars: Code for Performance Free Intel webinars can help you accelerate application performance. Explore tips for MPI, OpenMP, advanced profiling, and more. Get the most from the latest Intel processors and coprocessors. See abstracts and register http://pubads.g.doubleclick.net/gampad/clk?id=60134791iu=/4140/ostg.clktrk ___ Languagetool-devel mailing list Languagetool-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/languagetool-devel
Re: Modules for individual supported languages?
On 04.10.2013, at 10:03, Marcin Miłkowski list-addr...@wp.pl wrote: The only remaining problem is that for bilingual rules, we really need some mechanisms to communicate between the modules, and to download modules on the fly. Office users don't use that, but for CheckMate (translation QA) that could be a problem. Downloading on the fly could be solved by hooking into the resource loader mechanism that has been suggested elsewhere. In fact, I that would be implemented, I was thinking of using it for enabling exactly that. In DKPro Core, we enabled many of the language analysis modules to automatically download their models from a Maven repository, but for LanguageTool, we currently still bundle the whole bunch because this loader mechanism is lacking. -- Richard -- October Webinars: Code for Performance Free Intel webinars can help you accelerate application performance. Explore tips for MPI, OpenMP, advanced profiling, and more. Get the most from the latest Intel processors and coprocessors. See abstracts and register http://pubads.g.doubleclick.net/gampad/clk?id=60134791iu=/4140/ostg.clktrk ___ Languagetool-devel mailing list Languagetool-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/languagetool-devel
Re: Facebook stats
Might be just a coincidence. Anyway, recently, I've been planting baits on QA and bugs forums of OOo and LibO, plus other places. Hope to attract coders that way too. Jan Schreiber wrote thus at 04:40 PM 04-10-13: Hello everybody, FYI, here are a few screenshots that show recent statistics about our Facebook page. User activity there is increasing slowly but steadily. In the last few days, two users have contacted us via our Facebook wall. --Jan -- October Webinars: Code for Performance Free Intel webinars can help you accelerate application performance. Explore tips for MPI, OpenMP, advanced profiling, and more. Get the most from the latest Intel processors and coprocessors. See abstracts and register http://pubads.g.doubleclick.net/gampad/clk?id=60134791iu=/4140/ostg.clktrk ___ Languagetool-devel mailing list Languagetool-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/languagetool-devel
Re: He tried not to laughs.
By verb I mean of course the wrong verb (i.e., non-base form). Anyway, mails you from always end up as below: all joined up. Marcin MiÅkowski wrote thus at 04:04 PM 04-10-13: W dniu 2013-10-04 05:00, Kumara Bhikkhu pisze: Disambiguator. I don't even know what that is. Never mind. Thanks. I'll add rules for not to (verb) and (verb) to (verb), and see how that goes. What do you want to achieve this way? These will not be error patterns... Best, Marcin kb Marcin Mià kowski wrote thus at 06:39 PM 03-10-13: W dniu 2013-10-03 11:43, Kumara Bhikkhu pisze: Marcin Mià kowski wrote thus at 04 04:15 PM 03-10-13: Hi, W dniu 2013-10-03 06:24, Kumara Bhikkhu pisze: Can the one who created this contact me personally? It's not triggering He tried not to laugh_s_. I don't know how to correct it. I'll write this on the list -- laughs is also plural of laugh, which is excluded by the exception below (NNS). Unfortunately, without this exception, a lot of false alarms are found.I thought so.Now, maybe we could have a second variant of the rule that takes not +to, and then the exception would not be required. This would have to be tested on a large corpus. Anyway of indicating verbs but excepting those what are also nouns? No, it's not possible unless you have a perfect rule in the disambiguator for this. Best, Marcin -- October Webinars: Code for Performance Free Intel webinars can help you accelerate application performance. Explore tips for MPI, OpenMP, advanced profiling, and more. Get the most from the latest Intel processors and coprocessors. See abstracts and register http://pubads.g.doubleclick.net/gampad/clk?id=60134791iu=/4140/ostg.clktrk ___ Languagetool-devel mailing list Languagetool-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/languagetool-devel -- October Webinars: Code for Performance Free Intel webinars can help you accelerate application performance. Explore tips for MPI, OpenMP, advanced profiling, and more. Get the most from the latest Intel processors and coprocessors. See abstracts and register http://pubads.g.doubleclick.net/gampad/clk?id=60134791iu=/4140/ostg.clktrk ___ Languagetool-devel mailing list Languagetool-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/languagetool-devel . -- October Webinars: Code for Performance Free Intel webinars can help you accelerate application performance. Explore tips for MPI, OpenMP, advanced profiling, and more. Get the most from the latest Intel processors and coprocessors. See abstracts and register http://pubads.g.doubleclick.net/gampad/clk?id=60134791iu=/4140/ostg.clktrk ___ Languagetool-devel mailing list Languagetool-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/languagetool-devel -- October Webinars: Code for Performance Free Intel webinars can help you accelerate application performance. Explore tips for MPI, OpenMP, advanced profiling, and more. Get the most from the latest Intel processors and coprocessors. See abstracts and register http://pubads.g.doubleclick.net/gampad/clk?id=60134791iu=/4140/ostg.clktrk ___ Languagetool-devel mailing list Languagetool-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/languagetool-devel
Re: building a synthesizer
Daniel, I found the same problem recently. I resorted to the attached perl script for this step. Regards, Jaume Ortolà 2013/10/4 Daniel Naber list2...@danielnaber.de Hi, did anybody recently build a synthesizer? When I follow the instructions at http://wiki.languagetool.org/developing-a-tagger-dictionary#toc8 I get messages like this: Line number 1 has less than 3 tab-separated fields: |I So what's the correct input format for the tab2morph step? When I use a format with three tab-separated columns the synth dict I get is very large (10MB). Regards Daniel -- http://www.danielnaber.de -- October Webinars: Code for Performance Free Intel webinars can help you accelerate application performance. Explore tips for MPI, OpenMP, advanced profiling, and more. Get the most from the latest Intel processors and coprocessors. See abstracts and register http://pubads.g.doubleclick.net/gampad/clk?id=60134791iu=/4140/ostg.clktrk ___ Languagetool-devel mailing list Languagetool-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/languagetool-devel morph_data_ca.pl Description: Binary data -- October Webinars: Code for Performance Free Intel webinars can help you accelerate application performance. Explore tips for MPI, OpenMP, advanced profiling, and more. Get the most from the latest Intel processors and coprocessors. See abstracts and register http://pubads.g.doubleclick.net/gampad/clk?id=60134791iu=/4140/ostg.clktrk___ Languagetool-devel mailing list Languagetool-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/languagetool-devel
Re: building a synthesizer
Hi Why aren't the scripts in of all binary dictionaries in Git? They are useful when dictionaries need to be upgraded. And they help maintainers of other languages to figure out how to their dictionaries. A script is less ambiguous than documentation. I have not created a synthesizer dictionary yet, but the POS and FSA for French and Breton are created from scripts in Git: languagetool-language-modules/fr/src/main/resources/org/languagetool/resource/fr/create-lexicon.sh languagetool-language-modules/br/src/main/resources/org/languagetool/resource/br/create-lexicon.pl languagetool-language-modules/br/src/main/resources/org/languagetool/resource/br/hunspell/create-fsa-spell-dictionary.sh In fact, I saw that one of the reasons for not packaging LanguageTool in Debian, is because we don't automate the creation of binary dictionaries: https://bugs.launchpad.net/ubuntu/+source/openoffice.org/+bug/114375 Regards Dominique Jaume Ortolà i Font wrote: Daniel, I found the same problem recently. I resorted to the attached perl script for this step. Regards, Jaume Ortolà 2013/10/4 Daniel Naber list2...@danielnaber.de Hi, did anybody recently build a synthesizer? When I follow the instructions at http://wiki.languagetool.org/developing-a-tagger-dictionary#toc8 I get messages like this: Line number 1 has less than 3 tab-separated fields: |I So what's the correct input format for the tab2morph step? When I use a format with three tab-separated columns the synth dict I get is very large (10MB). Regards Daniel -- http://www.danielnaber.de -- October Webinars: Code for Performance Free Intel webinars can help you accelerate application performance. Explore tips for MPI, OpenMP, advanced profiling, and more. Get the most from the latest Intel processors and coprocessors. See abstracts and register http://pubads.g.doubleclick.net/gampad/clk?id=60134791iu=/4140/ostg.clktrk ___ Languagetool-devel mailing list Languagetool-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/languagetool-devel
Re: building a synthesizer
W dniu 2013-10-04 18:28, Daniel Naber pisze: Hi, did anybody recently build a synthesizer? When I follow the instructions at http://wiki.languagetool.org/developing-a-tagger-dictionary#toc8 I get messages like this: Line number 1 has less than 3 tab-separated fields: |I So what's the correct input format for the tab2morph step? When I use a format with three tab-separated columns the synth dict I get is very large (10MB). The message can be turned off, and should be turned off. This is just a warning from the tab2morph tool by using -nw switch. For example: java -jar $(morfologik) tab2morph -nw -i synt.txt -o synt_in.txt 2 /dev/null Best, Marcin -- October Webinars: Code for Performance Free Intel webinars can help you accelerate application performance. Explore tips for MPI, OpenMP, advanced profiling, and more. Get the most from the latest Intel processors and coprocessors. See abstracts and register http://pubads.g.doubleclick.net/gampad/clk?id=60134791iu=/4140/ostg.clktrk ___ Languagetool-devel mailing list Languagetool-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/languagetool-devel
Re: building a synthesizer
On 2013-10-04 19:28, Dominique Pellé wrote: Why aren't the scripts in of all binary dictionaries in Git? Why have scripts at all? Now with maven we can set up a small project that has morfologik-tools as a dependency and that does all import and export work by calling the morfologik methods from Java. Anyway, Marcin's hint to use the -nw option helped and I now have the synthesizer I needed. Regards Daniel -- http://www.danielnaber.de -- October Webinars: Code for Performance Free Intel webinars can help you accelerate application performance. Explore tips for MPI, OpenMP, advanced profiling, and more. Get the most from the latest Intel processors and coprocessors. See abstracts and register http://pubads.g.doubleclick.net/gampad/clk?id=60134791iu=/4140/ostg.clktrk ___ Languagetool-devel mailing list Languagetool-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/languagetool-devel
Re: xml editor
Daniel Naber wrote thus at 02:12 PM 03-10-13: On 2013-09-27 10:30, Kumara Bhikkhu wrote: I realise that working on an XML with a normal text editor is no fun. What's a good (free) xml editor for our purpose? This is only complementary to a local editor, but it helps to test the rule against real data: http://community.languagetool.org/ruleEditor/expert I wish it had code completion but the underlying editor doesn't support that yet (or not easily). I know this one. Wish it can do more than one rule at a time though. I was referring to an editor that can tell if I missed out an end tag (perhaps because I accidentally deleted it) from the xml. kb -- October Webinars: Code for Performance Free Intel webinars can help you accelerate application performance. Explore tips for MPI, OpenMP, advanced profiling, and more. Get the most from the latest Intel processors and coprocessors. See abstracts and register http://pubads.g.doubleclick.net/gampad/clk?id=60134791iu=/4140/ostg.clktrk ___ Languagetool-devel mailing list Languagetool-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/languagetool-devel
