Hi Sane,
Rada Mihalcea's alignment page
(http://www.cs.unt.edu/~rada/wa/#data) has links to human-aligned
alignment files, such as those used in the ACL 2005 workshop on parallel
text, the NAACL 2003 workshop on Parallel Text or the Europarl corpus
word alignment trial and test data. Note that the widely used
French-English Hansards reference contains many ambiguous (i.e. P) links
and thus favours high precision computed alignments, as explained in
"Patrik Lambert, Adrià de Gispert, Rafael Banchs and José B. Mariño.
Guidelines for Word Alignment Evaluation and Manual Alignment. Language
Resources and Evaluation, 39 (4) pp. 267-285. Springer"
Patrik
[EMAIL PROTECTED] wrote:
> Message: 2
> Date: Thu, 17 Jul 2008 15:10:29 +0200
> From: "Sanne Korzec" <[EMAIL PROTECTED]>
> Subject: [Moses-support] recaser
> To: <[email protected]>
> Message-ID:
> <[EMAIL PROTECTED]>
>
> 3) Finally and a little off topic, does anybody in our research field
> have word-based human-aligned alignment files for some test set available?
>
_______________________________________________
Moses-support mailing list
[email protected]
http://mailman.mit.edu/mailman/listinfo/moses-support
