diff options
| author | Patrick Simianer <simianer@cl.uni-heidelberg.de> | 2012-12-28 10:28:55 +0100 |
|---|---|---|
| committer | Patrick Simianer <simianer@cl.uni-heidelberg.de> | 2012-12-28 10:28:55 +0100 |
| commit | cf7d5f83d2fd5731bda7b1265775a395a5d3c1ba (patch) | |
| tree | 3223f3e7ffba5126e3efda46aadb4ca09c737718 /python/pkg/cdec/sa/extractor.py | |
| parent | f9ec8ec31e5a0b6ccc352c834733462eae2481a0 (diff) | |
| parent | a72b677092b44d35fec07cc2acf1064210c14ec5 (diff) | |
Merge remote-tracking branch 'upstream/master'
Diffstat (limited to 'python/pkg/cdec/sa/extractor.py')
| -rw-r--r-- | python/pkg/cdec/sa/extractor.py | 13 |
1 files changed, 12 insertions, 1 deletions
diff --git a/python/pkg/cdec/sa/extractor.py b/python/pkg/cdec/sa/extractor.py index e09f79ea..62a251a7 100644 --- a/python/pkg/cdec/sa/extractor.py +++ b/python/pkg/cdec/sa/extractor.py @@ -1,5 +1,5 @@ from itertools import chain -import os +import os, sys import cdec.configobj from cdec.sa.features import EgivenFCoherent, SampleCountF, CountEF,\ MaxLexEgivenF, MaxLexFgivenE, IsSingletonF, IsSingletonFE @@ -82,3 +82,14 @@ class GrammarExtractor: meta = cdec.sa.annotate(words) cnet = cdec.sa.make_lattice(words) return self.factory.input(cnet, meta) + + # Add training instance to data + def add_instance(self, sentence, reference, alignment): + f_words = cdec.sa.encode_words(sentence.split()) + e_words = cdec.sa.encode_words(reference.split()) + al = sorted(tuple(int(i) for i in pair.split('-')) for pair in alignment.split()) + self.factory.add_instance(f_words, e_words, al) + + # Debugging + def dump_online_stats(self): + self.factory.dump_online_stats()
\ No newline at end of file |