Running tt4j with other language configuration

[BogdanMaier]

Is there something implemeneted to run for instance the romanian configuration of TT? I understand we pass as paramter the TT source folder, but how do we controll all this attributes from the wrapper?

!/bin/sh

Set these paths appropriately

BIN=/Users/bogdan/private/resources/treetagger/bin CMD=/Users/bogdan/private/resources/treetagger/cmd LIB=/Users/bogdan/private/resources/treetagger/lib

OPTIONS="-token -lemma -sgml"

TOKENIZER=${CMD}/utf8-tokenize.perl TAGGER=${BIN}/tree-tagger ABBR_LIST=${LIB}/romanian-abbreviations PARFILE=${LIB}/romanian-utf8.par

$TOKENIZER -r -a $ABBR_LIST $* | ${CMD}/split-romanian.perl ${LIB}/romanian-tokens | $TAGGER $OPTIONS $PARFILE

Thanks, Bogdan

[reckart]

TT4J only covers the invocation of the actual binary. Tokenization is not supported by the wrapper. You have to do that yourself. Relevant options from your list would be

BIN=/Users/bogdan/private/resources/treetagger/bin
OPTIONS="-token -lemma -sgml"
TAGGER=${BIN}/tree-tagger
PARFILE=${LIB}/romanian-utf8.par
$TAGGER $OPTIONS $PARFILE

I have taken the example from the TT4J website here and added comments where your options fit in. Mind they do not fit in copy/paste, but I'm just pointing to related settings in TT4J.

package org.annolab.tt4j;

import static java.util.Arrays.asList;

public class Example {
  public static void main(String[] args) throws Exception {
    // Point TT4J to the TreeTagger installation directory. The executable is expected
    // in the "bin" subdirectory - in this example at "/opt/treetagger/bin/tree-tagger"
    System.setProperty("treetagger.home", "/opt/treetagger"); // <== options "BIN" and "TAGGER"
    TreeTaggerWrapper tt = new TreeTaggerWrapper<String>();
    try {
      tt.setModel("/opt/treetagger/models/english.par:iso8859-1"); // <== option "PARFILE"
      tt.setHandler(new TokenHandler<String>() {
        public void token(String token, String pos, String lemma) {
          System.out.println(token + "\t" + pos + "\t" + lemma);
        }
      });
      tt.process(asList(new String[] { "This", "is", "a", "test", "." }));
    }
    finally {
      tt.destroy();
    }
  }
}

The options -quiet -no-unknown -sgml -token -lemma are enabled by default. If you want to change that, you can call tt.setArguments(...). Mind that TT4J expects a certain output from TT, so the options should be at least -quiet -sgml -token -lemma.

See also: https://reckart.github.io/tt4j/usage.html

[BogdanMaier]

Thanks for the prompt response :)