hierarchical extraction

[Hieu Hoang]

Hi cdec'ers and Joshuaers

I was trying to figure out if the hiero extraction methods are the same across our 3 systems.

How do you guys create hierachical rules? Do you follow Chiang's algorithm and just poke holes in a phrase-based rule to get non-terminals? 

--
Hieu Hoang
Research Associate
University of Edinburgh
http://www.hoang.co.uk/hieu

[Hieu Hoang]

I was asking more about the algorithm but comparing the extracted output might also a good idea. It's a comparison of just the extract file, right, not the phrase-table?

Lets start by just comparing the extracted file from 10 aligned sentences.

I've put the sentences here:
   http://www.statmt.org/~s0565741/download/extract/

The datafile are: source, target, alignment

My extracted files are in the directories:

   extract/?/

On 17 February 2014 22:22, Matt Post <po...@cs.jhu.edu> wrote:

I'd be curious how similar they are. Joshua uses Thrax whose default parameters are lifted from David's paper.

If you want to compare, we could generate grammars on a fixed dataset and then share them.

Matt

--
You received this message because you are subscribed to the Google Groups "Joshua Technical Support" group.
To unsubscribe from this group and stop receiving emails from it, send an email to joshua_suppor...@googlegroups.com.
To post to this group, send email to joshua_...@googlegroups.com.
Visit this group at http://groups.google.com/group/joshua_support.
For more options, visit https://groups.google.com/groups/opt_out.

--
You received this message because you are subscribed to the Google Groups "Joshua Technical Support" group.
To unsubscribe from this group and stop receiving emails from it, send an email to joshua_suppor...@googlegroups.com.
To post to this group, send email to joshua_...@googlegroups.com.
Visit this group at http://groups.google.com/group/joshua_support.
For more options, visit https://groups.google.com/groups/opt_out.