Patrik Lambert > Lingua-AlignmentSet-1.1 > evaluate_alSet-version.pl

Download:
Lingua-AlignmentSet-1.1.tgz

Annotate this POD

View/Report Bugs
Source  

NAME ^

evaluate_alSet-version.pl - Evaluates submitted Alignment Set(s) against an answer Alignment Set

SYNOPSIS ^

perl evaluate_alSet-version.pl [options] required_arguments

Required arguments:

        -sub FILENAME,'DESCRIPTION'    As many as submission source-to-target links files.
        -subf BLINKER|GIZA|NAACL    Submission file(s) format (required if not TALP).
        -ans FILENAME    Answer source-to-target links file
        -ansf BLINKER|GIZA|NAACL    Answer file format (required if not TALP)

Options:

        -sub_range BEGIN-END    Submission Alignment Set range
        -ans_range BEGIN-END    Answer Alignment Set range
        -alignMode as-is|null-align|no-null-align Alignment mode. Default: no-null-align
        -w    Activates the weighting of the links
        -title Title of the experiment series
        -help|?    Prints the help and exits
        -man    Prints the manual and exits

ARGUMENTS ^

--sub,--submission FILENAME,'DESCRIPTION'

One entry for each submission source-to-target (i.e. links) file name (or directory, in case of BLINKER format). Optionally a description can be added, between '' if it contains white spaces.

--subf,--sub_format BLINKER|GIZA|NAACL

Submission Alignment Set format (required if different from default, TALP). The same format is required for all input files.

--ans,--answer FILENAME

Answer source-to-target (i.e. links) file name (or directory, in case of BLINKER format)

--ansf,--ans_format BLINKER|GIZA|NAACL

Answer Alignment Set format (required if different from default, TALP)

OPTIONS ^

--sub_range BEGIN-END

Range of the submission source-to-target file (BEGIN and END are the sentence pair numbers). The same range is required for all input files.

--ans_range BEGIN-END

Range of the answer source-to-target file (BEGIN and END are the sentence pair numbers)

--alignMode as-is|null-align|no-null-align

Take alignment "as-is" or force NULL alignment or NO-NULL alignment (see AlignmentSet.pm documentation). The default here is 'no-null-align' (as opposed to the other scripts, where the default is 'as-is'). Use "as-is" only if you are sure answer and submission files are in the same alignment mode.

-w, --weighted

Weights the links according to the number of links of each word in the sentence pair.

--title

Give a title to the table where results are compared

--help, --?

Prints a help message and exits.

--man

Prints a help message and exits.

DESCRIPTION ^

Evaluates one or various submitted Alignment Set(s) against an answer Alignment Set, and compare the results in a table.

EXAMPLES ^

perl evaluate_alSet-version.pl -sub test-giza.spa2eng.giza,'Spanish to English' -sub test-giza.eng2spa.giza,'English to Spanish' -title'Alignment Evaluation' -subf=GIZA -ans test-answer.spa2eng.naacl

Gives the following output:

    Alignment Evaluation   
----------------------------------
 Experiment                Ps     Rs      Fs      Pp      Rp      Fp     AER  

Spanish to English 93.95 67.51 78.57 93.95 67.51 78.57 21.43

English to Spanish 81.57 74.14 77.68 86.31 65.60 74.54 20.07

AUTHOR ^

Patrik Lambert <lambert@gps.tsc.upc.edu> Some code from Rada Mihalcea's wa_eval_align.pl (http:://www.cs.unt.edu/rada/wpt/code/) has been integrated in the library function.

COPYRIGHT AND LICENSE ^

Copyright 2004-2005 by Patrick Lambert

This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License (version 2 or any later version).

syntax highlighting: