Workshop

(start screen and audio recording)

Intro (Daniel, 5-10 minutes):

• describe basic idea: a rule system that describes errors in XML
• flowchart of LT

Workshop:

• 1. block (Daniel, 15 minutes)
  • download the current OXT
  • install it in OOo and try on English text
  • try the GUI: java -jar LanguageToolGUI.jar
  • mention command line and HTTP server

• 2. block (Marcin, 40 minutes)
  • show that the English rules are in in rules/en/grammar.xml
  • explain that using an XML editor is a good idea
  • explain the syntax of a simple rule - one of the English redundant phrases, this is simple but useful
  • create a rule to match "foo bar"
  • restart LT and try it (GUI version, not OOo)
  • run a test using testrules.sh/testrules.bat
  • mention that it's important to use real examples
  • explain a more complex rule
    • what is a POS tagger
      • mention SENT_START / SENT_END tags available to all languages
    • it_is - a couple of cases, one with POS tags, and a couple others (as a rulegroup)
    • both… and in English, as it requires skipping
    • a rule for repeated phrases (can be French or Dutch)
  • if time permits:
    • explain complex suggestions and testing corrections
    • what is a synthesizer

• 3. block (Marcin, 15 minutes)
  • show how we add a new language by adapting the Java code
    • cvs -d:pserver:ten.egrofecruos.svc.lootegaugnal|suomynona#ten.egrofecruos.svc.lootegaugnal|suomynona:/cvsroot/languagetool login [type return when asked for password]
    • cvs -z3 -d:pserver:ten.egrofecruos.svc.lootegaugnal|suomynona#ten.egrofecruos.svc.lootegaugnal|suomynona:/cvsroot/languagetool co -P JLanguageTool
    • cd JLanguageTool
    • ant
    • if all works well, a file called "dist/LanguageTool-1.0.0-dev.oxt" is generated which you can install in OOo (or unzip to use the command line LanguageTool)
  • see http://www.languagetool.org/development/#newlanguage
  • explain that people can usually start working in a different language, i.e. write their rules for a new language in the grammar.xml of an existing one

• 4. block (Daniel, 10 minutes)
  • explain community.languagetool.org website
    • how to use it to find false alarms in Wikipedia
    • how to find content for rules (Wikipedia typos page, community feedback…)
    • … TODO….

• 5. block: if time permits and if attendees are interested in Java development (Daniel, 30 minutes):
  • how to setup LanguageTool in Eclipse
    • check out from CVS (see above)
    • File -> Import…
    • General, Existing Projects into Workspace
    • select the "JLanguageTool" directory just checked out from CVS
    • you can ignore compile errors in the "dev" package (or add the dependency, or delete the affected *.java files), but you need to fix all other compile errors
  • structure of a rule written in Java
  • write your own rule
  • explain that sentence segmentation / disambiguation is also rule-driven
    • and that it influences the default "Uppercase sentence start" rule (Marcin)

Finally (Marcin, 5 minutes):

• mention resources: homepage, Wiki, mailing list

(stop screen and audio recording)