I have an unusual question. A customer recently lost a translation model to a hard disk failure. This included the original phrase table and its binarized copy. He found a backup but for some reason, he could only recover the phrase-table.binphr.srctree and phrase-table.binphr.tgt binary files. The phrase-table.gz, phrase-table.binphr.idx, phrase-table.binphr.srcvoc, and phrase-table.binphr.tgtvoc files are permanently gone. To make things worse, the original training corpus is also gone and he would have to rebuild it from scratch.
So, here's the question. Is it possible to rebuild the phrase-table.binphr.idx, phrase-table.binphr.srcvoc, and phrase-table.binphr.tgtvoc files files from the phrase-table.binphr.srctree and phrase-table.binphr.tgt binary files? I suspect not, but thought it's worth asking. Thanks.
_______________________________________________ Moses-support mailing list [email protected] http://mailman.mit.edu/mailman/listinfo/moses-support
