Only one alignment generated for words during unsupervised iterations

anoopkunchukuttan / transliterator

Unsupervised Transliterator using phonetic features (particularly for Indian languages)

GNU General Public License v3.0

5 stars 3 forks source link

Closed anoopkunchukuttan closed 9 years ago

anoopkunchukuttan commented 9 years ago

There is only one alignment per word in every iteration
The reason for one alignment per iteration is that the initial mappings support only 1-1 alignments due to which subsequent iterations also learn only 1-1 alignments.
This is a consequence of the fact that source and target alignments have the same length and hence only one alignment is possible - the 1-1 alignment
This will need resolution of bug #3 and possibly #2

anoopkunchukuttan commented 9 years ago

As discussed bug #3 was indeed the reason for this bug. Since bug #3 has been resolved, this issued has also been resolved. Verified the fix