diff --git a/src/metagger/megam_classifier.py.in b/src/metagger/megam_classifier.py.in index d1376ff0e3cb7e8a4ad4a160a46fa0cddbfa88ca..69b1635dff2ca26ec4524ffae5bccb18abfe4e2e 100755 --- a/src/metagger/megam_classifier.py.in +++ b/src/metagger/megam_classifier.py.in @@ -60,7 +60,7 @@ class MegamClassifier: -maxit: max # of iterations (megam default:100) """ - print ">>> Training Megam classifier..." + print >> sys.stderr, ">>> Training Megam classifier..." self.paramfile = paramfile # build process command diff --git a/src/metagger/utils.py.in b/src/metagger/utils.py.in index 4cd49173a34f5cca7512e3f3c489454b7b738814..ac12bcf1d33a65b1afe5e934b96b6227e22407c6 100644 --- a/src/metagger/utils.py.in +++ b/src/metagger/utils.py.in @@ -8,16 +8,14 @@ def tag_dict(file_path): for s in BrownReader(file_path): for wd,tag in s: tag_dict[wd] = tag_dict.get(wd,[]) + [tag] - infile.close() return tag_dict -def word_list(file_path): +def word_list(file_path,t=5): word_ct = {} for s in BrownReader(file_path): for wd,tag in s: word_ct[wd] = word_ct.get(wd,0) + 1 - infile.close() filtered_wd_list = {} for w in word_ct: ct = word_ct[w]