#9 Phrase-based model - Copy if you can

Speaker:
Ondřej Bojar
Abstract:
The ninth MT talk introduces the current dominant approach to machine translation, phrase-based model, as implemented in the open-source Moses decoder. The phrase-based model decomposes input sentences into short sequences of word forms, disregarding word formation or any syntax. Despite it's simplicity, it is surprisingly successful, especially in repetitive domains.
Length:
00:08:06
Date:
07/04/2015
views:

Images:
Attachments: (video, slides, etc.)
12 MB
139 downloads
67 MB
155 downloads
32 MB
145 downloads
24 MB
151 downloads