LanguageTools Transducer Compiler:
Installation and Use

The compiler tools are stand-alone applications. They read a data file and transform it into a finite-state transducer compiled form. The input file syntax is described here.

Delivered Files

  • README.html
  • doc/SYNTAX.html (Explains the input syntax)
  • TransCompiler (binary file containing the executable compiler)
  • TransCompTest.src (Test source data)
  • (Test feature table data)

Running the Compiler

To run the compiler you need a source file (file *.src), built using the syntax explained above. The file which will be compiled and converted into a transducer (file *.tra)

  • Compile the delivered source file (TransCompTest.src)
    ./TransCompiler TransCompTest.src TransCompTest.tra

Now you are able to verify the generated transducer using one of the respective LanguageTools products, for example a lemmatizer or a simple analyzer. Use your own application to test its content, or:

  • Edit the file, delivered with your LanguageTools product, in order to perform some of the following requests:
    Please note: XYZ in this file name will be either Ger, Eng, Ita or Fre, representing the language of your product.
  •     baute
  • Test the queries, for example using the analyzer:
    java -cp 
      .:analyzer.jar com.canoo.wmtrans.analyzer.XYZIntegrationDemo
    where TransCompTest.tra represents the test compiled data, and represents features table.
    (Please note that Unix and Windows use different classpath seperators. Unix => ':', Windows => ';')

Remark: If you need to load two different transducers (for example, an analyzer built on your own data and an analyzer with the data delivered by Canoo), use the newInstance(...) method to instantiate the different transducers.