Rooba
/
spaCy
mirror of https://github.com/explosion/spaCy.git
1
1
Fork 0
Code Issues Packages Projects Releases Wiki Activity

Improve negative example handling in NER oracle

This commit is contained in:
Matthew Honnibal 2017-07-20 00:18:49 +02:00
parent a231b56d40
commit 0ca5832427
1 changed files with 11 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

12
spacy/syntax/ner.pyx
View File

@ -127,6 +127,8 @@ cdef class BiluoPushDown(TransitionSystem):
if name == '-' or name == None: if name == '-' or name == None:
move_str = 'M' move_str = 'M'
label = 0 label = 0
elif name == '!O':
return Transition(clas=0, move=ISNT, label=0, score=0)
elif '-' in name: elif '-' in name:
move_str, label_str = name.split('-', 1) move_str, label_str = name.split('-', 1)
# Hacky way to denote 'not this entity' # Hacky way to denote 'not this entity'
@ -308,6 +310,9 @@ cdef class In:
elif g_act == UNIT: elif g_act == UNIT:
# I, Gold U --> True iff next tag == O # I, Gold U --> True iff next tag == O
return next_act != OUT return next_act != OUT
# Support partial supervision in the form of "not this label"
elif g_act == ISNT:
return 0
else: else:
return 1 return 1
@ -350,6 +355,9 @@ cdef class Last:
elif g_act == UNIT: elif g_act == UNIT:
# L, Gold U --> True # L, Gold U --> True
return 0 return 0
# Support partial supervision in the form of "not this label"
elif g_act == ISNT:
return 0
else: else:
return 1 return 1
@ -418,7 +426,9 @@ cdef class Out:
cdef int g_act = gold.ner[s.B(0)].move cdef int g_act = gold.ner[s.B(0)].move
cdef attr_t g_tag = gold.ner[s.B(0)].label cdef attr_t g_tag = gold.ner[s.B(0)].label
if g_act == MISSING or g_act == ISNT: if g_act == ISNT and g_tag == 0:
return 1
elif g_act == MISSING or g_act == ISNT:
return 0 return 0
elif g_act == BEGIN: elif g_act == BEGIN:
# O, Gold B --> False # O, Gold B --> False