Poll: What advanced text entry method(s) would you like to see on Sailfish?
Poll Options
What advanced text entry method(s) would you like to see on Sailfish?

Reply
Thread Tools
Posts: 105 | Thanked: 205 times | Joined on Dec 2015 @ Spain
#201
Good news. OKBoard already has Spanish language to work with

All the problems was related to amount of RAM. 4GB isn't enough to process corpus.txt of 3 GB, so I have to reduce it drastically. For next week I already have ordered 16Gb for my PC so I can improve the accuracy of Spanish predictive in OKBoard.

Thank you for all your support!! I'll keep you informed about the process of a RPM creation to install Spanish language for OKBoard
 

The Following 7 Users Say Thank You to ferlanero For This Useful Post:
Posts: 635 | Thanked: 1,535 times | Joined on Feb 2014 @ Germany
#202
I think that a good dictionary is far more important than an extremely huge corpora file. Mine is only 200mb compressed and with my dictionary the prediction accuracy is nearly perfect.
 

The Following User Says Thank You to mautz For This Useful Post:
Posts: 89 | Thanked: 243 times | Joined on Jun 2014
#203
Originally Posted by mautz View Post
I think that a good dictionary is far more important than an extremely huge corpora file. Mine is only 200mb compressed and with my dictionary the prediction accuracy is nearly perfect.
This probably depends on the language? With Finnish for example, you need to have all the inflected forms of a word (kirjoittaa-kirjoitan-kirjoitat-kirjoitamme-kirjoitatte-kirjoittavat and a few dozen more for the verb ”to write”, just to have the common ones) and you need to know which one is most probably used in a context...

(I would very much like to have OKBoard working with Finnish, but just haven't had the time to look into building it myself yet... Hoping someone will beat me to it
 

The Following User Says Thank You to ssahla For This Useful Post:
Posts: 102 | Thanked: 187 times | Joined on Jan 2010
#204
@ssahla, @mautz, yes, a good mix of genres and basic understanding of lexicon and the collected corpus data can probably save some time and effort. 100 MB compressed is probably the threshold to aim to get over for the corpus data. Zipfs law will make it hard to find more than a fraction of the inflection forms of morphologically rich languages, so a good frequency dictionary could be beneficial if it is hard to find available corpus data.
 

The Following User Says Thank You to ljo For This Useful Post:
Posts: 635 | Thanked: 1,535 times | Joined on Feb 2014 @ Germany
#205
I used all the news files from the leipzig website to build my corpus file. The wiki articles use a to formal and sometimes scientific speech for everday use i think. For the dictionary i found a file which contains the 10000 most used words of my language(german in my case) and i dumped my sms conversation from my jolla and added them to the dictionary file as well. A good think to add to the corpora file are some ebooks which use everyday language.

I'm fine tuning my dictionary and corpus file and will look into building an rpm file to release the german language on openrepos in a few days.
 

The Following 6 Users Say Thank You to mautz For This Useful Post:
Posts: 635 | Thanked: 1,535 times | Joined on Feb 2014 @ Germany
#206
German Language for OKBoard- first try. Feedback welcome.

https://www.dropbox.com/s/qwruts98l9...mv7hl.rpm?dl=0
 

The Following 8 Users Say Thank You to mautz For This Useful Post:
Posts: 105 | Thanked: 205 times | Joined on Dec 2015 @ Spain
#207
Originally Posted by mautz View Post
German Language for OKBoard- first try. Feedback welcome.

https://www.dropbox.com/s/qwruts98l9...mv7hl.rpm?dl=0
Hi, Mautz! I already have the files of Spanish working with OKBoard... so can you explain here the steps for building the .rpm for the distribution, please?
 

The Following 4 Users Say Thank You to ferlanero For This Useful Post:
Posts: 635 | Thanked: 1,535 times | Joined on Feb 2014 @ Germany
#208
I used Schturmans Guide for NOOBS.
It is not necessary to build the RPMs as root or use the root home directory.

Just modify the spec file for example like this:


Code:
Name:          okboard-spanish
Version:       0.1
Release:       1
Summary:       Spanish language for OKBoard
Group:         System/Tools
Vendor:        YourName/Alias
Distribution:  SailfishOS
Packager: YourName/Alias <your@mail.adress>


License:       GPL

%description
Spanish language files for OkBoard

%files

%defattr(-,root,root,-)
/usr/share/okboard/es.tre.gz
/usr/share/okboard/predict-es.db.gz
/usr/share/okboard/predict-es.ng.gz

%post

%postun
if [ $1 = 0 ]; then
    // Do stuff specific to uninstalls
rm -rf /usr/share/okboard/es.tre.gz
rm -rf /usr/share/okboard/predict-es.db.gz
rm -rf /usr/share/okboard/predict-es.ng.gz

else
if [ $1 = 1 ]; then
    // Do stuff specific to upgrades
echo "It's just upgrade"
fi
fi

%changelog
* Wed Jan 20 2016 YourName <your@mail.adress> 0.1
- First build
GZip the three files for OKBoard like described in the Readme.md.

Create the folder
/home/nemo/rpmbuild/BUILD/okboard-spanish-0.1-1.arm/usr/share/okboard

The folder name in the BUILD directory depends on the spec file entries.
<Name>-<Version>-<Release>.arm

Copy the three OKBoard language files into the newly created folder.

Copy your created .spec file to /home/nemo/rpmbuild/SPECS

Build the RPM on your Jolla.

Last edited by mautz; 2016-01-20 at 13:51.
 

The Following 4 Users Say Thank You to mautz For This Useful Post:
Posts: 44 | Thanked: 180 times | Joined on Nov 2014
#209
Originally Posted by mautz View Post
German Language for OKBoard- first try. Feedback welcome.

https://www.dropbox.com/s/qwruts98l9...mv7hl.rpm?dl=0
also bei mir klappt es schon ganz gut .einige, zum Teil wichtige Worte kennt das Wörterbuch (noch) nicht aber diesen Text tippe ich mit dem Okboard.
Danke fürs hoch laden!

EDIT:
Sorry for writing in german, I had to try the german dictionary for OKBoard

I just wrote some SMS and no longer texts yet, so this is my first Impression. BUt I can say: Looks very promising, only two issues which make usage a bit difficult:
  • Some common words, which I would expect in the list of most used words, seem to be missing. E.g. "Danke" (=thank you) isn't recognized (nor is it in the suggestion list).
  • Words with Umlauts are rarely recognised. I guess that this feature is not implemented in OKBoard yet.

Thanks for your work @Eber, @mautz and everybody else!

EDIT2:
to enter words with umlauts you have to swipe over the non-umlauted letter (eg Ol for Öl and uber for über)
so no problems with umlauts ...

Last edited by LameDuck; 2016-01-20 at 16:28.
 

The Following User Says Thank You to LameDuck For This Useful Post:
Posts: 738 | Thanked: 819 times | Joined on Jan 2012 @ Berlin
#210
Originally Posted by mautz View Post
German Language for OKBoard- first try. Feedback welcome.

https://www.dropbox.com/s/qwruts98l9...mv7hl.rpm?dl=0
some basic stuf like "deiner" is not working perfekt. but thank you very match !!!! now i can use swype like keyboard ^^

now with color smileys from dolphin keyboard and all is good enough
__________________
www.sailfishmods.de
 
Reply

Tags
bettertxtentry, huntnpeck sucks, okboard, sailfish, swype


 
Forum Jump


All times are GMT. The time now is 11:01.