/workflow-docalignment

The "Document Alignment Tool Wrapper" of the ACCURAT toolkit.

Primary LanguagePerlApache License 2.0Apache-2.0

This directory contains all the resources and the tools that compose the "Document Alignment Workflow" described in the ACCURAT deliverable D2.6. Version 3.
The present bundle is compiled ONLY FOR Windows systems (!) and contains the following tools:

Requirements:
- 'perl' and 'java' must be in the %PATH%.
- IT IS STRONGLY RECOMMENDED that for EMACC, the user edits the cluster.info  (see D2.6 for that) file and run on multiple CPU cores since the algorithm is CPU intesive.
- be sure to use the environment variable USER to your user in cmd.exe! That is, 'echo %USER%' MUST NOT BE VOID! GIZA++ breaks if this does not happen.

1. CTS ComMetric "document aligner" application (ComMetric). Files belonging to this application are (directory 'commetric\'):
	- ComMetric.jar
	- en-stopwords.txt

2. CTS DictMetric "document aligner" application (DicMetric). Files belonging to this application are (directory 'dicmetric\'):
	- DicMetric.jar
	- 'dict\' directory
	- 'stopwords\' directory

6. RACAI "document aligner" application (EMACC). Files belonging to this application are (directory 'emacc-pexacc-lexacc\'):
	- emacc2.pl, precompworker.pl, emaccconf.pm, hddmatrix.pm
	- cluster.info
	- 'dict\' and 'res\' directories

7. USFD "document aligner" application (Feature-based Document pair classifier). Files belonging to this application are:
	- all files in the 'featclass\' directory.
	- featclass.bat

Files in the root of the arhive that MUST NOT be deleted:
	- en-stopwords.txt
	- DocumentAlignment.*
	- all directories

In order to run the workflow, please edit ONLY the 'DocumentAlignment.prop' for configuration and run 'DocumentAlignment.pl'.
Get usage information of 'DocumentAlignment.pl' by executing it without command line arguments.

IMPORTANT: Do not delete ANY files other than those produced by 'DocumentAlignment.pl' !!