AlvisNLP

corpus processing engine

Getting started

Using the command-line

In order to use AlvisNLP you need to Download and install it.

AlvisNLP consists on a single command-line executable: alvisnlp. In order to get a short help on the options:

$ alvisnlp -help

alvisnlp
    process text corpora

Usage:
    alvisnlp [OPTIONS] PLANFILE

Documentation options:
    -help                                 print this help
    -supportedModules                     print supported modules and exit
    -supportedConverters                  print all types of parameters that can be converted
    -supportedLibraries                   print the name of all supported expression libraries
    -moduleDoc              MODULE        print documentation for a module and exit
    -converterDoc           TYPE          print documentation for converters to a parameter type and exit
    -libraryDoc             LIB           print documentation for library LIB and exit
    -locale                 LOCALE        set locale for documentation and messages
    -supportedModulesXML                  print supported modules in XML and exit
    -supportedConvertersXML               print all types of parameters that can be converted in XML and exit
    -supportedLibrariesXML                print the name of all supported expression libraries in XML and exit
    -moduleDocXML           MODULE        print documentation for a module in XML and exit
    -converterDocXML        TYPE          print documentation for converter to a parameter type in XML and exit
    -libraryDocXML          LIB           print documentation for library LIB in XML and exit
    -planDoc                              write plan documentation instead of processing
    -docTransformer         XSLT          use the specified XSL transformation file to display documentation
    -xslParam               NAME VALUE    pass parameter to the style sheet specified by -docTransformer

Verbosity options:
    -verbose           log more information
    -quiet             log less information
    -silent            log only warnings
    -log     FILE      write log into FILE
    -append            append log at the end of the log file (ignore if no -log)
    -locale  LOCALE    set locale for documentation and messages

Plan options:
    -param                  MODULE PARAM VALUE    set to VALUE the parameter PARAM of module MODULE
    -xparam                 MODULE XML_PARAM      set MODULE parameter value specified by XML_PARAM
    -alias                  ALIAS VALUE           set to VALUE the parameter alias ALIAS
    -xalias                 XML_ALIAS             set parameter alias value specified by XML_ALIAS
    -unset                  MODULE PARAM          unset the parameter PARAM of module MODULE
    -defaultParamValuesFile FILE                  specifies the file containing default parameter values
    -module                 ID CLASS              append a module with the specified id and class at the end of the plan
    -shell                                        runs a shell after processing the corpus
    -creator                FEAT                  set the name of the feature containing the module that created each element
    -feat                   KEY VALUE             set feature for the corpus
    -entity                 NAME REPLACEMENT      define an XML entity replacement used in the plan file
    -environmentEntities                          define an XML entity replacement for each environment variable used in the plan file
    -propEntities           PROPFILE              define XML entity replacements from a properties file
    -inputDir               DIR                   add default input directory
    -outputDir              DIR                   set root output directory

Resume mode and dump options:
    -resume     FILE           resume processing from a dumped corpus
    -dumpModule MODULE FILE    set the dump file after the specified module
    -nodumps                   ignore all corpus dumps

Other options:
    -version          print version and exit
    -noProcess        do not process the corpus
    -writePlan        write the plan to standard output, adding type to the parameter aliases, using the XSLT provided by -docTransformer
    -tmp       DIR    set root of temporary directories
    -cleanTmp         delete the temp directory after processing
    -noColors         do not use ANSI color escape codes for logging or documentation

There are a lot of options because AlvisNLP is very versatile. The most important is the PLANFILE.

Plan file

You specify the sequence of modules with an XML file called the plan file. The plan file contains:

Here’s an example of a simple plan file that tells AlvisNLP to read text files, look for a specific regular expression pattern and writes the result as HTML so we can take a look at it.

example.plan

<alvisnlp-plan id="example">
    <read class="TextFileReader">
        <sourcePath>/path/to/txt/files</sourcePath>
    </read>

    <regexp class="RegExp">
        <pattern>\b[A-Z]\S*\b</pattern>
    <targetLayerName>capitalized</targetLayerName>
    </regexp>

    <write class="QuickHTML">
        <outDir>/path/to/output/dir</outDir>
    <classFeature>_dummy</classFeature>
    </write>
</alvisnlp-plan>

In this example, we define a plan of three successive steps named read, regexp, and write. These steps use the modules TextFileReader, RegExp, and QuickHTML respectively.

Within each step, we set some parameter values:

  1. For the step read, there’s a single parameter sourcePath that specifies where to read text files.
  2. For regexp, there are two parameters. pattern specifies the regular expression to search. targetLayerName is the name of the container where regexp will store matches. Here we call it “capitalized”.

Ready to run this plan:

$ alvisnlp example.plan

Further reading