Skip to content
Snippets Groups Projects
Select Git revision
  • 89b3540c5b6cad2531b2d9623bf28a93e12ff58a
  • main default protected
  • change_data_model
  • feature/add_auth_asr_service
  • fix/incorrect_import
  • feature/change_registry_clarin
  • feature/add_base_asr_service
  • feature/add_poetry
  • feature/add_word_ids
  • feature/add_sziszapangma
10 results

README.rst

Blame
  • README 2.21 KiB
    IKAR comes with a couple of useful scripts which make it possible to divide 
    resolution process into steps and select only specified part of interest.
    
    cdIkar.py <dir> - Classifies all .xml files in given directory
    convIkar.py <dir> - Converts all classified files in given directory into semEval format. All .btch_relpos are treated as model and .rel_cls as classified files. 
    cvIkar.py <dir> - Performs cross validation of given folder based on 'config' file. It is recommended to launch 'eval' after folds are classified to see semEval measures
    daemonIkar <dir> - Daemon version of ikar script. Loads models only once. Requires previous launch of 'goDaemons' script
    exIkar.py <dir> - Exports all classified files from given folder into inforex transfer file. Refined .rel_r files are treated as classification outcome. Export is saved in a single export.ex file.
    goDaemons - Launches ikar daemon
    killDaemons - Stops ikar daemon
    ikar <dir> - Performs full anaphora resolution task on all .txt files in given directory. Requires all prequisites to run.
    impIkar.py <dumpFile> <database> <user> <password> - Imports an Inforex transfer file into given inforex database. The database needs to already have documents with ids matching those in transer file.
    pcIkar.py <dir> -Performs plain classification of directory. Unlike cdIkar it creates an own relation file. Doesn't only base on already existing rel.xml file as a base for classification.
    pmIkar.py - Prepares classifier models based on anp/ directory. idClassifier and ptClassifier are created
    rIkar.py <dir> - Refines all .rel_cls files in given directory, selecting only one relation from each annotation 
    semiikar <dir> - Performs anaphora resolution task on all .xml files in given directory. The files must be ccl's with proper names annotated.
    
    REQUIRED TO RUN:
    (It is recommended to install them in given order)
    -Corpus2
    -Toki
    -Morfeusz
    -Maca
    -WMBT
    (Configuration of Nerd&Liner might take some time and patience (optionally tranquilizers)
    -Nerd
    -Liner2
    -Fextor
    -LexCSD (Modular) Also set pythonPath to each of modules in .bashrc file.
    -IOBBER
    
    INSTALLATION:
    There is no installation script at this time.
    To make all scripts run properly you need to set proper paths in config files:
    -config
    -settings