Éva Mújdricza-Maydt

Gutenberg corpus

This slightly differing version of the Bilingual Formal / Informal Address Corpus was used to train and test the CRF-based sentence aligner CRFalign.

• training set: 111* file pairs (80.6 MB)
• test set: 3 file pairs, with golden annotation (420 KB)
• training and test set: 114* file pairs (81.0 MB)
• readme

(* NOTE that one file pair in our training set cannot be provided due to copyright restrictions. See readme.)