diff --git a/semantics/management/commands/frame_compare_0.py b/semantics/management/commands/frame_compare_0.py
index cf2f871..986e255 100644
--- a/semantics/management/commands/frame_compare_0.py
+++ b/semantics/management/commands/frame_compare_0.py
@@ -25,21 +25,25 @@ class Command(BaseCommand):
def handle(self, **options):
# SPLIT FRAMES IN BUNLES
+ print "splitting frames in bundles..."
frames = SemanticFrame.objects.filter(next=None, removed=False, entry__isnull=False).order_by('id')
frame_bundles = []
frame_bundle = []
i = 0 # frame_bundle size
for frame in frames:
- frame_bundle.append(frame)
- i += 1
- if i == BUNDLE_SIZE:
- frame_bundles.append(frame_bundle)
- frame_bundle = []
- i = 0
+ if frame.entry.actual_lemma().status.priority >= 90: # (S)ready / (S)validated
+ lus = frame.lexical_units.all()
+ if len(lus) > 0 and\
+ max([lu.luid for lu in lus]) >=0:
+ frame_bundle.append(frame)
+ i += 1
+ if i == BUNDLE_SIZE:
+ frame_bundles.append(frame_bundle)
+ frame_bundle = []
+ i = 0
if i != 0:
frame_bundles.append(frame_bundle)
-
- print len(frames)
+ print " ...done"
# LOCAL DATABASE CONNECTION
session, TT_dict = get_db_data()
diff --git a/semantics/management/commands/frame_compare_1.py b/semantics/management/commands/frame_compare_1.py
index 9a6808f..986e255 100644
--- a/semantics/management/commands/frame_compare_1.py
+++ b/semantics/management/commands/frame_compare_1.py
@@ -17,7 +17,7 @@ from settings import PROJECT_PATH
BUNDLE_SIZE = 50
PROCESSES_NO = 16
-PROCESS_ID = 1
+PROCESS_ID = 0
class Command(BaseCommand):
args = 'none'
@@ -25,20 +25,26 @@ class Command(BaseCommand):
def handle(self, **options):
# SPLIT FRAMES IN BUNLES
+ print "splitting frames in bundles..."
frames = SemanticFrame.objects.filter(next=None, removed=False, entry__isnull=False).order_by('id')
frame_bundles = []
frame_bundle = []
i = 0 # frame_bundle size
for frame in frames:
- frame_bundle.append(frame)
- i += 1
- if i == BUNDLE_SIZE:
- frame_bundles.append(frame_bundle)
- frame_bundle = []
- i = 0
+ if frame.entry.actual_lemma().status.priority >= 90: # (S)ready / (S)validated
+ lus = frame.lexical_units.all()
+ if len(lus) > 0 and\
+ max([lu.luid for lu in lus]) >=0:
+ frame_bundle.append(frame)
+ i += 1
+ if i == BUNDLE_SIZE:
+ frame_bundles.append(frame_bundle)
+ frame_bundle = []
+ i = 0
if i != 0:
frame_bundles.append(frame_bundle)
-
+ print " ...done"
+
# LOCAL DATABASE CONNECTION
session, TT_dict = get_db_data()
LexicalUnit._session = session
@@ -54,13 +60,13 @@ class Command(BaseCommand):
f = Frame.from_slowal(frame)
frames1.append(f)
if i == j:
- match_frames_diagonal(frames1, session, TT_dict) # , verbose=True, fake=True)
+ match_frames_diagonal(frames1, session, TT_dict)# , verbose=True, fake=True)
else:
frames2 = []
for frame in frame_bundles[j]:
f = Frame.from_slowal(frame)
frames2.append(f)
- match_frames(frames1, frames2, session, TT_dict) # , verbose=True, fake=True)
+ match_frames(frames1, frames2, session, TT_dict)# , verbose=True, fake=True)
c += 1
diff --git a/semantics/management/commands/frame_compare_10.py b/semantics/management/commands/frame_compare_10.py
index 28c65e2..986e255 100644
--- a/semantics/management/commands/frame_compare_10.py
+++ b/semantics/management/commands/frame_compare_10.py
@@ -17,7 +17,7 @@ from settings import PROJECT_PATH
BUNDLE_SIZE = 50
PROCESSES_NO = 16
-PROCESS_ID = 10
+PROCESS_ID = 0
class Command(BaseCommand):
args = 'none'
@@ -25,20 +25,26 @@ class Command(BaseCommand):
def handle(self, **options):
# SPLIT FRAMES IN BUNLES
+ print "splitting frames in bundles..."
frames = SemanticFrame.objects.filter(next=None, removed=False, entry__isnull=False).order_by('id')
frame_bundles = []
frame_bundle = []
i = 0 # frame_bundle size
for frame in frames:
- frame_bundle.append(frame)
- i += 1
- if i == BUNDLE_SIZE:
- frame_bundles.append(frame_bundle)
- frame_bundle = []
- i = 0
+ if frame.entry.actual_lemma().status.priority >= 90: # (S)ready / (S)validated
+ lus = frame.lexical_units.all()
+ if len(lus) > 0 and\
+ max([lu.luid for lu in lus]) >=0:
+ frame_bundle.append(frame)
+ i += 1
+ if i == BUNDLE_SIZE:
+ frame_bundles.append(frame_bundle)
+ frame_bundle = []
+ i = 0
if i != 0:
frame_bundles.append(frame_bundle)
-
+ print " ...done"
+
# LOCAL DATABASE CONNECTION
session, TT_dict = get_db_data()
LexicalUnit._session = session
@@ -54,13 +60,13 @@ class Command(BaseCommand):
f = Frame.from_slowal(frame)
frames1.append(f)
if i == j:
- match_frames_diagonal(frames1, session, TT_dict) # , verbose=True, fake=True)
+ match_frames_diagonal(frames1, session, TT_dict)# , verbose=True, fake=True)
else:
frames2 = []
for frame in frame_bundles[j]:
f = Frame.from_slowal(frame)
frames2.append(f)
- match_frames(frames1, frames2, session, TT_dict) # , verbose=True, fake=True)
+ match_frames(frames1, frames2, session, TT_dict)# , verbose=True, fake=True)
c += 1
diff --git a/semantics/management/commands/frame_compare_11.py b/semantics/management/commands/frame_compare_11.py
index 75e8125..986e255 100644
--- a/semantics/management/commands/frame_compare_11.py
+++ b/semantics/management/commands/frame_compare_11.py
@@ -17,7 +17,7 @@ from settings import PROJECT_PATH
BUNDLE_SIZE = 50
PROCESSES_NO = 16
-PROCESS_ID = 11
+PROCESS_ID = 0
class Command(BaseCommand):
args = 'none'
@@ -25,20 +25,26 @@ class Command(BaseCommand):
def handle(self, **options):
# SPLIT FRAMES IN BUNLES
+ print "splitting frames in bundles..."
frames = SemanticFrame.objects.filter(next=None, removed=False, entry__isnull=False).order_by('id')
frame_bundles = []
frame_bundle = []
i = 0 # frame_bundle size
for frame in frames:
- frame_bundle.append(frame)
- i += 1
- if i == BUNDLE_SIZE:
- frame_bundles.append(frame_bundle)
- frame_bundle = []
- i = 0
+ if frame.entry.actual_lemma().status.priority >= 90: # (S)ready / (S)validated
+ lus = frame.lexical_units.all()
+ if len(lus) > 0 and\
+ max([lu.luid for lu in lus]) >=0:
+ frame_bundle.append(frame)
+ i += 1
+ if i == BUNDLE_SIZE:
+ frame_bundles.append(frame_bundle)
+ frame_bundle = []
+ i = 0
if i != 0:
frame_bundles.append(frame_bundle)
-
+ print " ...done"
+
# LOCAL DATABASE CONNECTION
session, TT_dict = get_db_data()
LexicalUnit._session = session
@@ -54,13 +60,13 @@ class Command(BaseCommand):
f = Frame.from_slowal(frame)
frames1.append(f)
if i == j:
- match_frames_diagonal(frames1, session, TT_dict) # , verbose=True, fake=True)
+ match_frames_diagonal(frames1, session, TT_dict)# , verbose=True, fake=True)
else:
frames2 = []
for frame in frame_bundles[j]:
f = Frame.from_slowal(frame)
frames2.append(f)
- match_frames(frames1, frames2, session, TT_dict) # , verbose=True, fake=True)
+ match_frames(frames1, frames2, session, TT_dict)# , verbose=True, fake=True)
c += 1
diff --git a/semantics/management/commands/frame_compare_12.py b/semantics/management/commands/frame_compare_12.py
index e51d4f5..986e255 100644
--- a/semantics/management/commands/frame_compare_12.py
+++ b/semantics/management/commands/frame_compare_12.py
@@ -17,7 +17,7 @@ from settings import PROJECT_PATH
BUNDLE_SIZE = 50
PROCESSES_NO = 16
-PROCESS_ID = 12
+PROCESS_ID = 0
class Command(BaseCommand):
args = 'none'
@@ -25,20 +25,26 @@ class Command(BaseCommand):
def handle(self, **options):
# SPLIT FRAMES IN BUNLES
+ print "splitting frames in bundles..."
frames = SemanticFrame.objects.filter(next=None, removed=False, entry__isnull=False).order_by('id')
frame_bundles = []
frame_bundle = []
i = 0 # frame_bundle size
for frame in frames:
- frame_bundle.append(frame)
- i += 1
- if i == BUNDLE_SIZE:
- frame_bundles.append(frame_bundle)
- frame_bundle = []
- i = 0
+ if frame.entry.actual_lemma().status.priority >= 90: # (S)ready / (S)validated
+ lus = frame.lexical_units.all()
+ if len(lus) > 0 and\
+ max([lu.luid for lu in lus]) >=0:
+ frame_bundle.append(frame)
+ i += 1
+ if i == BUNDLE_SIZE:
+ frame_bundles.append(frame_bundle)
+ frame_bundle = []
+ i = 0
if i != 0:
frame_bundles.append(frame_bundle)
-
+ print " ...done"
+
# LOCAL DATABASE CONNECTION
session, TT_dict = get_db_data()
LexicalUnit._session = session
@@ -54,13 +60,13 @@ class Command(BaseCommand):
f = Frame.from_slowal(frame)
frames1.append(f)
if i == j:
- match_frames_diagonal(frames1, session, TT_dict) # , verbose=True, fake=True)
+ match_frames_diagonal(frames1, session, TT_dict)# , verbose=True, fake=True)
else:
frames2 = []
for frame in frame_bundles[j]:
f = Frame.from_slowal(frame)
frames2.append(f)
- match_frames(frames1, frames2, session, TT_dict) # , verbose=True, fake=True)
+ match_frames(frames1, frames2, session, TT_dict)# , verbose=True, fake=True)
c += 1
diff --git a/semantics/management/commands/frame_compare_13.py b/semantics/management/commands/frame_compare_13.py
index fbb8bfb..986e255 100644
--- a/semantics/management/commands/frame_compare_13.py
+++ b/semantics/management/commands/frame_compare_13.py
@@ -17,7 +17,7 @@ from settings import PROJECT_PATH
BUNDLE_SIZE = 50
PROCESSES_NO = 16
-PROCESS_ID = 13
+PROCESS_ID = 0
class Command(BaseCommand):
args = 'none'
@@ -25,20 +25,26 @@ class Command(BaseCommand):
def handle(self, **options):
# SPLIT FRAMES IN BUNLES
+ print "splitting frames in bundles..."
frames = SemanticFrame.objects.filter(next=None, removed=False, entry__isnull=False).order_by('id')
frame_bundles = []
frame_bundle = []
i = 0 # frame_bundle size
for frame in frames:
- frame_bundle.append(frame)
- i += 1
- if i == BUNDLE_SIZE:
- frame_bundles.append(frame_bundle)
- frame_bundle = []
- i = 0
+ if frame.entry.actual_lemma().status.priority >= 90: # (S)ready / (S)validated
+ lus = frame.lexical_units.all()
+ if len(lus) > 0 and\
+ max([lu.luid for lu in lus]) >=0:
+ frame_bundle.append(frame)
+ i += 1
+ if i == BUNDLE_SIZE:
+ frame_bundles.append(frame_bundle)
+ frame_bundle = []
+ i = 0
if i != 0:
frame_bundles.append(frame_bundle)
-
+ print " ...done"
+
# LOCAL DATABASE CONNECTION
session, TT_dict = get_db_data()
LexicalUnit._session = session
@@ -54,13 +60,13 @@ class Command(BaseCommand):
f = Frame.from_slowal(frame)
frames1.append(f)
if i == j:
- match_frames_diagonal(frames1, session, TT_dict) # , verbose=True, fake=True)
+ match_frames_diagonal(frames1, session, TT_dict)# , verbose=True, fake=True)
else:
frames2 = []
for frame in frame_bundles[j]:
f = Frame.from_slowal(frame)
frames2.append(f)
- match_frames(frames1, frames2, session, TT_dict) # , verbose=True, fake=True)
+ match_frames(frames1, frames2, session, TT_dict)# , verbose=True, fake=True)
c += 1
diff --git a/semantics/management/commands/frame_compare_14.py b/semantics/management/commands/frame_compare_14.py
index 5cde5ab..986e255 100644
--- a/semantics/management/commands/frame_compare_14.py
+++ b/semantics/management/commands/frame_compare_14.py
@@ -17,7 +17,7 @@ from settings import PROJECT_PATH
BUNDLE_SIZE = 50
PROCESSES_NO = 16
-PROCESS_ID = 14
+PROCESS_ID = 0
class Command(BaseCommand):
args = 'none'
@@ -25,20 +25,26 @@ class Command(BaseCommand):
def handle(self, **options):
# SPLIT FRAMES IN BUNLES
+ print "splitting frames in bundles..."
frames = SemanticFrame.objects.filter(next=None, removed=False, entry__isnull=False).order_by('id')
frame_bundles = []
frame_bundle = []
i = 0 # frame_bundle size
for frame in frames:
- frame_bundle.append(frame)
- i += 1
- if i == BUNDLE_SIZE:
- frame_bundles.append(frame_bundle)
- frame_bundle = []
- i = 0
+ if frame.entry.actual_lemma().status.priority >= 90: # (S)ready / (S)validated
+ lus = frame.lexical_units.all()
+ if len(lus) > 0 and\
+ max([lu.luid for lu in lus]) >=0:
+ frame_bundle.append(frame)
+ i += 1
+ if i == BUNDLE_SIZE:
+ frame_bundles.append(frame_bundle)
+ frame_bundle = []
+ i = 0
if i != 0:
frame_bundles.append(frame_bundle)
-
+ print " ...done"
+
# LOCAL DATABASE CONNECTION
session, TT_dict = get_db_data()
LexicalUnit._session = session
@@ -54,13 +60,13 @@ class Command(BaseCommand):
f = Frame.from_slowal(frame)
frames1.append(f)
if i == j:
- match_frames_diagonal(frames1, session, TT_dict) # , verbose=True, fake=True)
+ match_frames_diagonal(frames1, session, TT_dict)# , verbose=True, fake=True)
else:
frames2 = []
for frame in frame_bundles[j]:
f = Frame.from_slowal(frame)
frames2.append(f)
- match_frames(frames1, frames2, session, TT_dict) # , verbose=True, fake=True)
+ match_frames(frames1, frames2, session, TT_dict)# , verbose=True, fake=True)
c += 1
diff --git a/semantics/management/commands/frame_compare_15.py b/semantics/management/commands/frame_compare_15.py
index cb7f241..986e255 100644
--- a/semantics/management/commands/frame_compare_15.py
+++ b/semantics/management/commands/frame_compare_15.py
@@ -17,7 +17,7 @@ from settings import PROJECT_PATH
BUNDLE_SIZE = 50
PROCESSES_NO = 16
-PROCESS_ID = 15
+PROCESS_ID = 0
class Command(BaseCommand):
args = 'none'
@@ -25,20 +25,26 @@ class Command(BaseCommand):
def handle(self, **options):
# SPLIT FRAMES IN BUNLES
+ print "splitting frames in bundles..."
frames = SemanticFrame.objects.filter(next=None, removed=False, entry__isnull=False).order_by('id')
frame_bundles = []
frame_bundle = []
i = 0 # frame_bundle size
for frame in frames:
- frame_bundle.append(frame)
- i += 1
- if i == BUNDLE_SIZE:
- frame_bundles.append(frame_bundle)
- frame_bundle = []
- i = 0
+ if frame.entry.actual_lemma().status.priority >= 90: # (S)ready / (S)validated
+ lus = frame.lexical_units.all()
+ if len(lus) > 0 and\
+ max([lu.luid for lu in lus]) >=0:
+ frame_bundle.append(frame)
+ i += 1
+ if i == BUNDLE_SIZE:
+ frame_bundles.append(frame_bundle)
+ frame_bundle = []
+ i = 0
if i != 0:
frame_bundles.append(frame_bundle)
-
+ print " ...done"
+
# LOCAL DATABASE CONNECTION
session, TT_dict = get_db_data()
LexicalUnit._session = session
@@ -54,13 +60,13 @@ class Command(BaseCommand):
f = Frame.from_slowal(frame)
frames1.append(f)
if i == j:
- match_frames_diagonal(frames1, session, TT_dict) # , verbose=True, fake=True)
+ match_frames_diagonal(frames1, session, TT_dict)# , verbose=True, fake=True)
else:
frames2 = []
for frame in frame_bundles[j]:
f = Frame.from_slowal(frame)
frames2.append(f)
- match_frames(frames1, frames2, session, TT_dict) # , verbose=True, fake=True)
+ match_frames(frames1, frames2, session, TT_dict)# , verbose=True, fake=True)
c += 1
diff --git a/semantics/management/commands/frame_compare_2.py b/semantics/management/commands/frame_compare_2.py
index 755959b..986e255 100644
--- a/semantics/management/commands/frame_compare_2.py
+++ b/semantics/management/commands/frame_compare_2.py
@@ -17,7 +17,7 @@ from settings import PROJECT_PATH
BUNDLE_SIZE = 50
PROCESSES_NO = 16
-PROCESS_ID = 2
+PROCESS_ID = 0
class Command(BaseCommand):
args = 'none'
@@ -25,20 +25,26 @@ class Command(BaseCommand):
def handle(self, **options):
# SPLIT FRAMES IN BUNLES
+ print "splitting frames in bundles..."
frames = SemanticFrame.objects.filter(next=None, removed=False, entry__isnull=False).order_by('id')
frame_bundles = []
frame_bundle = []
i = 0 # frame_bundle size
for frame in frames:
- frame_bundle.append(frame)
- i += 1
- if i == BUNDLE_SIZE:
- frame_bundles.append(frame_bundle)
- frame_bundle = []
- i = 0
+ if frame.entry.actual_lemma().status.priority >= 90: # (S)ready / (S)validated
+ lus = frame.lexical_units.all()
+ if len(lus) > 0 and\
+ max([lu.luid for lu in lus]) >=0:
+ frame_bundle.append(frame)
+ i += 1
+ if i == BUNDLE_SIZE:
+ frame_bundles.append(frame_bundle)
+ frame_bundle = []
+ i = 0
if i != 0:
frame_bundles.append(frame_bundle)
-
+ print " ...done"
+
# LOCAL DATABASE CONNECTION
session, TT_dict = get_db_data()
LexicalUnit._session = session
@@ -54,13 +60,13 @@ class Command(BaseCommand):
f = Frame.from_slowal(frame)
frames1.append(f)
if i == j:
- match_frames_diagonal(frames1, session, TT_dict) # , verbose=True, fake=True)
+ match_frames_diagonal(frames1, session, TT_dict)# , verbose=True, fake=True)
else:
frames2 = []
for frame in frame_bundles[j]:
f = Frame.from_slowal(frame)
frames2.append(f)
- match_frames(frames1, frames2, session, TT_dict) # , verbose=True, fake=True)
+ match_frames(frames1, frames2, session, TT_dict)# , verbose=True, fake=True)
c += 1
diff --git a/semantics/management/commands/frame_compare_3.py b/semantics/management/commands/frame_compare_3.py
index 4b4dbda..986e255 100644
--- a/semantics/management/commands/frame_compare_3.py
+++ b/semantics/management/commands/frame_compare_3.py
@@ -17,7 +17,7 @@ from settings import PROJECT_PATH
BUNDLE_SIZE = 50
PROCESSES_NO = 16
-PROCESS_ID = 3
+PROCESS_ID = 0
class Command(BaseCommand):
args = 'none'
@@ -25,20 +25,26 @@ class Command(BaseCommand):
def handle(self, **options):
# SPLIT FRAMES IN BUNLES
+ print "splitting frames in bundles..."
frames = SemanticFrame.objects.filter(next=None, removed=False, entry__isnull=False).order_by('id')
frame_bundles = []
frame_bundle = []
i = 0 # frame_bundle size
for frame in frames:
- frame_bundle.append(frame)
- i += 1
- if i == BUNDLE_SIZE:
- frame_bundles.append(frame_bundle)
- frame_bundle = []
- i = 0
+ if frame.entry.actual_lemma().status.priority >= 90: # (S)ready / (S)validated
+ lus = frame.lexical_units.all()
+ if len(lus) > 0 and\
+ max([lu.luid for lu in lus]) >=0:
+ frame_bundle.append(frame)
+ i += 1
+ if i == BUNDLE_SIZE:
+ frame_bundles.append(frame_bundle)
+ frame_bundle = []
+ i = 0
if i != 0:
frame_bundles.append(frame_bundle)
-
+ print " ...done"
+
# LOCAL DATABASE CONNECTION
session, TT_dict = get_db_data()
LexicalUnit._session = session
@@ -54,13 +60,13 @@ class Command(BaseCommand):
f = Frame.from_slowal(frame)
frames1.append(f)
if i == j:
- match_frames_diagonal(frames1, session, TT_dict) # , verbose=True, fake=True)
+ match_frames_diagonal(frames1, session, TT_dict)# , verbose=True, fake=True)
else:
frames2 = []
for frame in frame_bundles[j]:
f = Frame.from_slowal(frame)
frames2.append(f)
- match_frames(frames1, frames2, session, TT_dict) # , verbose=True, fake=True)
+ match_frames(frames1, frames2, session, TT_dict)# , verbose=True, fake=True)
c += 1
diff --git a/semantics/management/commands/frame_compare_4.py b/semantics/management/commands/frame_compare_4.py
index 8ecd164..986e255 100644
--- a/semantics/management/commands/frame_compare_4.py
+++ b/semantics/management/commands/frame_compare_4.py
@@ -17,7 +17,7 @@ from settings import PROJECT_PATH
BUNDLE_SIZE = 50
PROCESSES_NO = 16
-PROCESS_ID = 4
+PROCESS_ID = 0
class Command(BaseCommand):
args = 'none'
@@ -25,20 +25,26 @@ class Command(BaseCommand):
def handle(self, **options):
# SPLIT FRAMES IN BUNLES
+ print "splitting frames in bundles..."
frames = SemanticFrame.objects.filter(next=None, removed=False, entry__isnull=False).order_by('id')
frame_bundles = []
frame_bundle = []
i = 0 # frame_bundle size
for frame in frames:
- frame_bundle.append(frame)
- i += 1
- if i == BUNDLE_SIZE:
- frame_bundles.append(frame_bundle)
- frame_bundle = []
- i = 0
+ if frame.entry.actual_lemma().status.priority >= 90: # (S)ready / (S)validated
+ lus = frame.lexical_units.all()
+ if len(lus) > 0 and\
+ max([lu.luid for lu in lus]) >=0:
+ frame_bundle.append(frame)
+ i += 1
+ if i == BUNDLE_SIZE:
+ frame_bundles.append(frame_bundle)
+ frame_bundle = []
+ i = 0
if i != 0:
frame_bundles.append(frame_bundle)
-
+ print " ...done"
+
# LOCAL DATABASE CONNECTION
session, TT_dict = get_db_data()
LexicalUnit._session = session
@@ -54,13 +60,13 @@ class Command(BaseCommand):
f = Frame.from_slowal(frame)
frames1.append(f)
if i == j:
- match_frames_diagonal(frames1, session, TT_dict) # , verbose=True, fake=True)
+ match_frames_diagonal(frames1, session, TT_dict)# , verbose=True, fake=True)
else:
frames2 = []
for frame in frame_bundles[j]:
f = Frame.from_slowal(frame)
frames2.append(f)
- match_frames(frames1, frames2, session, TT_dict) # , verbose=True, fake=True)
+ match_frames(frames1, frames2, session, TT_dict)# , verbose=True, fake=True)
c += 1
diff --git a/semantics/management/commands/frame_compare_5.py b/semantics/management/commands/frame_compare_5.py
index 853c558..986e255 100644
--- a/semantics/management/commands/frame_compare_5.py
+++ b/semantics/management/commands/frame_compare_5.py
@@ -17,7 +17,7 @@ from settings import PROJECT_PATH
BUNDLE_SIZE = 50
PROCESSES_NO = 16
-PROCESS_ID = 5
+PROCESS_ID = 0
class Command(BaseCommand):
args = 'none'
@@ -25,20 +25,26 @@ class Command(BaseCommand):
def handle(self, **options):
# SPLIT FRAMES IN BUNLES
+ print "splitting frames in bundles..."
frames = SemanticFrame.objects.filter(next=None, removed=False, entry__isnull=False).order_by('id')
frame_bundles = []
frame_bundle = []
i = 0 # frame_bundle size
for frame in frames:
- frame_bundle.append(frame)
- i += 1
- if i == BUNDLE_SIZE:
- frame_bundles.append(frame_bundle)
- frame_bundle = []
- i = 0
+ if frame.entry.actual_lemma().status.priority >= 90: # (S)ready / (S)validated
+ lus = frame.lexical_units.all()
+ if len(lus) > 0 and\
+ max([lu.luid for lu in lus]) >=0:
+ frame_bundle.append(frame)
+ i += 1
+ if i == BUNDLE_SIZE:
+ frame_bundles.append(frame_bundle)
+ frame_bundle = []
+ i = 0
if i != 0:
frame_bundles.append(frame_bundle)
-
+ print " ...done"
+
# LOCAL DATABASE CONNECTION
session, TT_dict = get_db_data()
LexicalUnit._session = session
@@ -54,13 +60,13 @@ class Command(BaseCommand):
f = Frame.from_slowal(frame)
frames1.append(f)
if i == j:
- match_frames_diagonal(frames1, session, TT_dict) # , verbose=True, fake=True)
+ match_frames_diagonal(frames1, session, TT_dict)# , verbose=True, fake=True)
else:
frames2 = []
for frame in frame_bundles[j]:
f = Frame.from_slowal(frame)
frames2.append(f)
- match_frames(frames1, frames2, session, TT_dict) # , verbose=True, fake=True)
+ match_frames(frames1, frames2, session, TT_dict)# , verbose=True, fake=True)
c += 1
diff --git a/semantics/management/commands/frame_compare_6.py b/semantics/management/commands/frame_compare_6.py
index 5e2d414..986e255 100644
--- a/semantics/management/commands/frame_compare_6.py
+++ b/semantics/management/commands/frame_compare_6.py
@@ -17,7 +17,7 @@ from settings import PROJECT_PATH
BUNDLE_SIZE = 50
PROCESSES_NO = 16
-PROCESS_ID = 6
+PROCESS_ID = 0
class Command(BaseCommand):
args = 'none'
@@ -25,20 +25,26 @@ class Command(BaseCommand):
def handle(self, **options):
# SPLIT FRAMES IN BUNLES
+ print "splitting frames in bundles..."
frames = SemanticFrame.objects.filter(next=None, removed=False, entry__isnull=False).order_by('id')
frame_bundles = []
frame_bundle = []
i = 0 # frame_bundle size
for frame in frames:
- frame_bundle.append(frame)
- i += 1
- if i == BUNDLE_SIZE:
- frame_bundles.append(frame_bundle)
- frame_bundle = []
- i = 0
+ if frame.entry.actual_lemma().status.priority >= 90: # (S)ready / (S)validated
+ lus = frame.lexical_units.all()
+ if len(lus) > 0 and\
+ max([lu.luid for lu in lus]) >=0:
+ frame_bundle.append(frame)
+ i += 1
+ if i == BUNDLE_SIZE:
+ frame_bundles.append(frame_bundle)
+ frame_bundle = []
+ i = 0
if i != 0:
frame_bundles.append(frame_bundle)
-
+ print " ...done"
+
# LOCAL DATABASE CONNECTION
session, TT_dict = get_db_data()
LexicalUnit._session = session
@@ -54,13 +60,13 @@ class Command(BaseCommand):
f = Frame.from_slowal(frame)
frames1.append(f)
if i == j:
- match_frames_diagonal(frames1, session, TT_dict) # , verbose=True, fake=True)
+ match_frames_diagonal(frames1, session, TT_dict)# , verbose=True, fake=True)
else:
frames2 = []
for frame in frame_bundles[j]:
f = Frame.from_slowal(frame)
frames2.append(f)
- match_frames(frames1, frames2, session, TT_dict) # , verbose=True, fake=True)
+ match_frames(frames1, frames2, session, TT_dict)# , verbose=True, fake=True)
c += 1
diff --git a/semantics/management/commands/frame_compare_7.py b/semantics/management/commands/frame_compare_7.py
index cd9f691..986e255 100644
--- a/semantics/management/commands/frame_compare_7.py
+++ b/semantics/management/commands/frame_compare_7.py
@@ -17,7 +17,7 @@ from settings import PROJECT_PATH
BUNDLE_SIZE = 50
PROCESSES_NO = 16
-PROCESS_ID = 7
+PROCESS_ID = 0
class Command(BaseCommand):
args = 'none'
@@ -25,20 +25,26 @@ class Command(BaseCommand):
def handle(self, **options):
# SPLIT FRAMES IN BUNLES
+ print "splitting frames in bundles..."
frames = SemanticFrame.objects.filter(next=None, removed=False, entry__isnull=False).order_by('id')
frame_bundles = []
frame_bundle = []
i = 0 # frame_bundle size
for frame in frames:
- frame_bundle.append(frame)
- i += 1
- if i == BUNDLE_SIZE:
- frame_bundles.append(frame_bundle)
- frame_bundle = []
- i = 0
+ if frame.entry.actual_lemma().status.priority >= 90: # (S)ready / (S)validated
+ lus = frame.lexical_units.all()
+ if len(lus) > 0 and\
+ max([lu.luid for lu in lus]) >=0:
+ frame_bundle.append(frame)
+ i += 1
+ if i == BUNDLE_SIZE:
+ frame_bundles.append(frame_bundle)
+ frame_bundle = []
+ i = 0
if i != 0:
frame_bundles.append(frame_bundle)
-
+ print " ...done"
+
# LOCAL DATABASE CONNECTION
session, TT_dict = get_db_data()
LexicalUnit._session = session
@@ -54,13 +60,13 @@ class Command(BaseCommand):
f = Frame.from_slowal(frame)
frames1.append(f)
if i == j:
- match_frames_diagonal(frames1, session, TT_dict) # , verbose=True, fake=True)
+ match_frames_diagonal(frames1, session, TT_dict)# , verbose=True, fake=True)
else:
frames2 = []
for frame in frame_bundles[j]:
f = Frame.from_slowal(frame)
frames2.append(f)
- match_frames(frames1, frames2, session, TT_dict) # , verbose=True, fake=True)
+ match_frames(frames1, frames2, session, TT_dict)# , verbose=True, fake=True)
c += 1
diff --git a/semantics/management/commands/frame_compare_8.py b/semantics/management/commands/frame_compare_8.py
index 91b854c..986e255 100644
--- a/semantics/management/commands/frame_compare_8.py
+++ b/semantics/management/commands/frame_compare_8.py
@@ -17,7 +17,7 @@ from settings import PROJECT_PATH
BUNDLE_SIZE = 50
PROCESSES_NO = 16
-PROCESS_ID = 8
+PROCESS_ID = 0
class Command(BaseCommand):
args = 'none'
@@ -25,20 +25,26 @@ class Command(BaseCommand):
def handle(self, **options):
# SPLIT FRAMES IN BUNLES
+ print "splitting frames in bundles..."
frames = SemanticFrame.objects.filter(next=None, removed=False, entry__isnull=False).order_by('id')
frame_bundles = []
frame_bundle = []
i = 0 # frame_bundle size
for frame in frames:
- frame_bundle.append(frame)
- i += 1
- if i == BUNDLE_SIZE:
- frame_bundles.append(frame_bundle)
- frame_bundle = []
- i = 0
+ if frame.entry.actual_lemma().status.priority >= 90: # (S)ready / (S)validated
+ lus = frame.lexical_units.all()
+ if len(lus) > 0 and\
+ max([lu.luid for lu in lus]) >=0:
+ frame_bundle.append(frame)
+ i += 1
+ if i == BUNDLE_SIZE:
+ frame_bundles.append(frame_bundle)
+ frame_bundle = []
+ i = 0
if i != 0:
frame_bundles.append(frame_bundle)
-
+ print " ...done"
+
# LOCAL DATABASE CONNECTION
session, TT_dict = get_db_data()
LexicalUnit._session = session
@@ -54,13 +60,13 @@ class Command(BaseCommand):
f = Frame.from_slowal(frame)
frames1.append(f)
if i == j:
- match_frames_diagonal(frames1, session, TT_dict) # , verbose=True, fake=True)
+ match_frames_diagonal(frames1, session, TT_dict)# , verbose=True, fake=True)
else:
frames2 = []
for frame in frame_bundles[j]:
f = Frame.from_slowal(frame)
frames2.append(f)
- match_frames(frames1, frames2, session, TT_dict) # , verbose=True, fake=True)
+ match_frames(frames1, frames2, session, TT_dict)# , verbose=True, fake=True)
c += 1
diff --git a/semantics/management/commands/frame_compare_9.py b/semantics/management/commands/frame_compare_9.py
index 6a5d30d..986e255 100644
--- a/semantics/management/commands/frame_compare_9.py
+++ b/semantics/management/commands/frame_compare_9.py
@@ -17,7 +17,7 @@ from settings import PROJECT_PATH
BUNDLE_SIZE = 50
PROCESSES_NO = 16
-PROCESS_ID = 9
+PROCESS_ID = 0
class Command(BaseCommand):
args = 'none'
@@ -25,20 +25,26 @@ class Command(BaseCommand):
def handle(self, **options):
# SPLIT FRAMES IN BUNLES
+ print "splitting frames in bundles..."
frames = SemanticFrame.objects.filter(next=None, removed=False, entry__isnull=False).order_by('id')
frame_bundles = []
frame_bundle = []
i = 0 # frame_bundle size
for frame in frames:
- frame_bundle.append(frame)
- i += 1
- if i == BUNDLE_SIZE:
- frame_bundles.append(frame_bundle)
- frame_bundle = []
- i = 0
+ if frame.entry.actual_lemma().status.priority >= 90: # (S)ready / (S)validated
+ lus = frame.lexical_units.all()
+ if len(lus) > 0 and\
+ max([lu.luid for lu in lus]) >=0:
+ frame_bundle.append(frame)
+ i += 1
+ if i == BUNDLE_SIZE:
+ frame_bundles.append(frame_bundle)
+ frame_bundle = []
+ i = 0
if i != 0:
frame_bundles.append(frame_bundle)
-
+ print " ...done"
+
# LOCAL DATABASE CONNECTION
session, TT_dict = get_db_data()
LexicalUnit._session = session
@@ -54,13 +60,13 @@ class Command(BaseCommand):
f = Frame.from_slowal(frame)
frames1.append(f)
if i == j:
- match_frames_diagonal(frames1, session, TT_dict) # , verbose=True, fake=True)
+ match_frames_diagonal(frames1, session, TT_dict)# , verbose=True, fake=True)
else:
frames2 = []
for frame in frame_bundles[j]:
f = Frame.from_slowal(frame)
frames2.append(f)
- match_frames(frames1, frames2, session, TT_dict) # , verbose=True, fake=True)
+ match_frames(frames1, frames2, session, TT_dict)# , verbose=True, fake=True)
c += 1
diff --git a/semantics/management/commands/frame_compare_modules/match_frames.py b/semantics/management/commands/frame_compare_modules/match_frames.py
index def8477..d365223 100644
--- a/semantics/management/commands/frame_compare_modules/match_frames.py
+++ b/semantics/management/commands/frame_compare_modules/match_frames.py
@@ -36,6 +36,15 @@ cut_value[u'Attribute'] = -np.log(0.8)
cut_value[u'Measure'] = -np.log(0.8)
cut_value[u'Lemma'] = -np.log(1.0)
+arg_rank = defaultdict(lambda: 0)
+arg_rank[u'Time'] = 1
+arg_rank[u'Location'] = 1
+arg_rank[u'Path'] = 1
+arg_rank[u'Attribute'] = 1
+arg_rank[u'Measure'] = 1
+arg_rank[u'Lemma'] = 1
+
+
def find_max_arg_matching_value(label, arglist1, arglist2, selprefs_table):
s1 = len(arglist1)
s2 = len(arglist2)
@@ -47,13 +56,14 @@ def find_max_arg_matching_value(label, arglist1, arglist2, selprefs_table):
cut = s2
ans_pos = hungarian_algorithm(array.copy())
result = 0
+ missing = [0, 0, 0]
for i, j in ans_pos:
# i and j are matched
if i < cut:
result += array[i][j]
else:
- result += cut_value[label]
- return result
+ missing[arg_rank[label]] += 1
+ return result, tuple(missing)
def match_undefined_preferences_and_max_match_the_rest(label, arglist1, arglist2, selprefs_table):
selprefs1 = copy(arglist1)
@@ -74,18 +84,35 @@ def match_undefined_preferences_and_max_match_the_rest(label, arglist1, arglist2
del selprefs2[j]
# i and j are matched
if len(selprefs1) == 0 and len(selprefs2) == 0:
- return 1.0
+ return 1.0, (0, 0, 0)
else:
return find_max_arg_matching_value(label, selprefs1, selprefs2, selprefs_table)
-
-
+
+
+misses_coefficient = defaultdict(lambda: -np.log(0.1))
+misses_coefficient[0.0] = -np.log(1.0)
+misses_coefficient[1.0/3] = -np.log(0.99)
+misses_coefficient[2.0/3] = -np.log(0.97)
+misses_coefficient[1.0] = -np.log(0.95)
+misses_coefficient[4.0/3] = -np.log(0.92)
+misses_coefficient[5.0/3] = -np.log(0.9)
+misses_coefficient[2.0] = -np.log(0.5)
+
def find_matching_value(frame1, frame2, selprefs_table):
labels = set(frame1.get_role_labels()) | set(frame2.get_role_labels())
tmp = 0
+ missing0 = 0
+ missing1 = 0
+ missing2 = 0
for label in sorted(labels):
- # tmp += find_max_arg_matching_value(label, frame1.get_arguments(label), frame2.get_arguments(label), selprefs_table)
- tmp += match_undefined_preferences_and_max_match_the_rest(label, frame1.get_arguments(label), frame2.get_arguments(label), selprefs_table)
- return np.exp(-tmp)
+ val, (m0, m1, m2) = match_undefined_preferences_and_max_match_the_rest(label, frame1.get_arguments(label), frame2.get_arguments(label), selprefs_table)
+ tmp += val
+ missing0 += m0
+ missing1 += m1
+ missing2 += m2
+ m = missing0 + (missing1 * 1.0) / 3
+ res = tmp + misses_coefficient[m]
+ return np.exp(-res)
def match_transformed_frames(frame1, frame2, rule, selprefs_table):
v = find_matching_value(frame1, frame2, selprefs_table)