Michel Weimerskirch » Archive

Sneak Peek

michel — Tue, 05 Sep 2006 20:09:58 +0000

The development of my grammar and style checker for the Luxembourgish language is coming along nicely. As you can see on the screenshot below, the OpenOffice.org integration is mostly finished. Read on for more information.

Although I do not excessively work on it all day long, I expect to release a first version during october. This version will be able to correct the “Eifeler Regel” as well as a few style rules.

As I have not made a final decision about the license of the source code yet, I will not release it along with the binary version. Instead, it might or might not be available at a later time.

Screenshot of the current development version. As most of the work has been done “under the hood”, you won’t see anything spectacular.

Grammar and style checker for the Luxembourgish language

michel — Mon, 26 Jun 2006 15:25:27 +0000

Over the weekend I started writing what will be the first ever developed grammar and style checker for the Luxembourgish language. Unfortunately exams at university start in about a month, so I won’t really be able to continue working on it till the beginning of August. Read on for a little more about what I’ve done already and what I planned to do.

I already completed a sentence and word tokenizer that splits a text into sentences and words, considering special cases such as dates or abbreviations (which contain period characters, so not necessarily every period character ends a sentence). Besides that, I wrote a prototype for a part of speech tagger (which attributes grammatical information to the tokens returned from the tokenizer) and I implemented a few basic grammar and style rules.

As it seems, Luxembourgish is particularly easy to check with computer software, so I hopefully will be able to create a rock solid product that not only helps native speakers but also allows non-native speakers to better understand the peculiarities of the Luxembourgish language.

The license of the program will probably be the LGPL. It will be written in Java so that it can easily have multiple interfaces like a simple text-only interface, an advanced graphical interface, an interface integrated into OpenOffice.org as well as a web interface using Java Server Pages.

Screenshot of the “proof of concept”-like program

Inside Spellchecker.lu

michel — Thu, 04 May 2006 22:08:34 +0000

I finally completed writing the toolset that I wrote for spellchecker.lu. Those tools parse text files to create text corpora, classify new words using statistical methods, detect common spelling mistakes, generate and upload the dictionary files (as well as the soon-to-be-released debian packages) and they allow me to control and maintain the online checker. All of this is now as simple as it could possibly be. For all of you that always wanted to get a glimpse of what’s going on “behind the scenes” of spellchecker.lu, I prepared a few screenshots of the graphical user interface.

Though this might or might not look fancy to you, the toolset does a damn fascinating job:

This is the starting point of the toolset, giving an overview of all the subroutines.

This picture shows a more enduring process.
For the tech-savvy of you: The toolset is actually a Bash script using dialog for the GUI. The subroutines are implemented with Java, AWK as well as some SED.

Spellchecker.lu finally online

michel — Sat, 22 Apr 2006 10:41:10 +0000

As of today, spellchecker.lu is online. It features a Luxembourgish spellchecking dictionary for OpenOffice.org, an additional plugin that corrects a special rule which hunspell can’t handle, as well as an “online checker” based on hunspell.

PS: The site is in Luxembourgish, but international content will follow soon.

Spellchecker.lu: Coming soon

michel — Thu, 29 Dec 2005 15:51:26 +0000

Yesterday evening I set up an announcement page for the spellchecking-project I am currently working on with friends of mine. Check out spellchecker.lu. Official release date will be the 2nd of February 2006. The countdown is running…