def look_up(self, anaphor, antecedent):
""" Look up strings of the mentions in the pair list.
Args:
anaphor (Mention): A mention.
antecedent (Mention): Another mention, the candidate antecedent
for anaphor.
Returns:
True if the pair of strings corresponding to anaphor of
antecedent, stripped determiners and possessive s, can be found
in the list of pairs.
"""
# whole string
anaphor_cleaned = " ".join(
util.clean_via_pos(anaphor.attributes["tokens"],
anaphor.attributes["pos"]))
antecedent_cleaned = " ".join(
util.clean_via_pos(antecedent.attributes["tokens"],
antecedent.attributes["pos"]))
return (
(anaphor_cleaned, antecedent_cleaned) in self.pairs
or (antecedent_cleaned, anaphor_cleaned) in self.pairs
)
def non_pronominal_string_match(anaphor, antecedent):
if anaphor.attributes["type"] in ["PRO", "DEM", "VRB"]:
return False
elif antecedent.attributes["type"] in ["PRO", "DEM", "VRB"]:
return False
else:
return (" ".join(util.clean_via_pos(anaphor.attributes["tokens"],
anaphor.attributes["pos"])).lower()
== " ".join(util.clean_via_pos(
antecedent.attributes["tokens"],
antecedent.attributes["pos"])).lower())
def non_pronominal_string_match(anaphor, antecedent):
if anaphor.attributes["type"] in ["PRO", "DEM", "VRB"]:
return False
elif antecedent.attributes["type"] in ["PRO", "DEM", "VRB"]:
return False
else:
return (" ".join(
util.clean_via_pos(anaphor.attributes["tokens"],
anaphor.attributes["pos"])).lower() == " ".join(
util.clean_via_pos(
antecedent.attributes["tokens"],
antecedent.attributes["pos"])).lower())
def not_compatible(anaphor, antecedent):
if (" ".join(
util.clean_via_pos(anaphor.attributes["tokens"],
anaphor.attributes["pos"])).lower() == " ".join(
util.clean_via_pos(
antecedent.attributes["tokens"],
antecedent.attributes["pos"])).lower()):
return False
gender = (anaphor.attributes["gender"] == "UNKNOWN"
or antecedent.attributes["gender"] == "UNKNOWN" or
anaphor.attributes["gender"] == antecedent.attributes["gender"])
number = (anaphor.attributes["number"] == "UNKNOWN"
or antecedent.attributes["number"] == "UNKNOWN" or
anaphor.attributes["number"] == antecedent.attributes["number"])
semantic_class = (anaphor.attributes["semantic_class"] == "UNKNOWN"
or antecedent.attributes["semantic_class"] == "UNKNOWN"
or anaphor.attributes["semantic_class"]
== antecedent.attributes["semantic_class"])
return not (gender and number and semantic_class)
def not_compatible(anaphor, antecedent):
if (" ".join(util.clean_via_pos(anaphor.attributes["tokens"],
anaphor.attributes["pos"])).lower() ==
" ".join(util.clean_via_pos(antecedent.attributes["tokens"],
antecedent.attributes["pos"])).lower()):
return False
gender = (anaphor.attributes["gender"] == "UNKNOWN"
or antecedent.attributes["gender"] == "UNKNOWN"
or anaphor.attributes["gender"]
== antecedent.attributes["gender"])
number = (anaphor.attributes["number"] == "UNKNOWN"
or antecedent.attributes["number"] == "UNKNOWN"
or anaphor.attributes["number"]
== antecedent.attributes["number"])
semantic_class = (anaphor.attributes["semantic_class"] == "UNKNOWN"
or antecedent.attributes["semantic_class"] == "UNKNOWN"
or anaphor.attributes["semantic_class"]
== antecedent.attributes["semantic_class"])
return not (gender and number and semantic_class)
def look_up(self, anaphor, antecedent):
""" Look up strings of the mentions in the pair list.
Args:
anaphor (Mention): A mention.
antecedent (Mention): Another mention, the candidate antecedent
for anaphor.
Returns:
True if the pair of strings corresponding to anaphor of
antecedent, stripped determiners and possessive s, can be found
in the list of pairs.
"""
# whole string
anaphor_cleaned = " ".join(
util.clean_via_pos(anaphor.attributes["tokens"],
anaphor.attributes["pos"]))
antecedent_cleaned = " ".join(
util.clean_via_pos(antecedent.attributes["tokens"],
antecedent.attributes["pos"]))
return ((anaphor_cleaned, antecedent_cleaned) in self.pairs
or (antecedent_cleaned, anaphor_cleaned) in self.pairs)
def substring(anaphor, antecedent):
if anaphor.attributes["type"] in ["PRO", "DEM", "VRB"]:
return False
elif antecedent.attributes["type"] != "NAM":
return False
elif (anaphor.attributes["semantic_class"] == "NUMERIC"
or antecedent.attributes["semantic_class"] == "NUMERIC"):
return False
elif anaphor.attributes["head"] == ["and"]:
return False
else:
cleaned = util.clean_via_pos(anaphor.attributes["tokens"],
anaphor.attributes["pos"])
return (" ".join(cleaned).lower()
in " ".join(antecedent.attributes["tokens"]).lower())
def substring(anaphor, antecedent):
if anaphor.attributes["type"] in ["PRO", "DEM", "VRB"]:
return False
elif antecedent.attributes["type"] != "NAM":
return False
elif (anaphor.attributes["semantic_class"] == "NUMERIC" or
antecedent.attributes["semantic_class"] == "NUMERIC"):
return False
elif anaphor.attributes["head"] == ["and"]:
return False
else:
cleaned = util.clean_via_pos(
anaphor.attributes["tokens"],
anaphor.attributes["pos"])
return (" ".join(cleaned).lower()
in " ".join(antecedent.attributes["tokens"]).lower())
def test_clean_via_pos(self):
self.assertEqual(
["newly-elect", "leader", "wife"],
clean_via_pos(["the", "newly-elect", "leader", "'s", "wife"],
["DT", "JJ", "NN", "POS", "NN"]))
def test_clean_via_pos(self):
self.assertEqual(
["newly-elect", "leader", "wife"],
clean_via_pos(
["the", "newly-elect", "leader", "'s", "wife"],
["DT", "JJ", "NN", "POS", "NN"]))
