-
Notifications
You must be signed in to change notification settings - Fork 36
/
Copy pathdueefin_post_process.py
181 lines (163 loc) · 6.88 KB
/
dueefin_post_process.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
import json
from collections import Counter, defaultdict
from itertools import combinations
from typing import DefaultDict, List, Optional
from dee.event_types import get_event_template
template = get_event_template("dueefin_wo_tgg")
event_type_fields_list = template.event_type_fields_list
event_type2event_class = template.event_type2event_class
def load_line_json(filepath):
data = []
with open(filepath, "rt", encoding="utf-8") as fin:
for line in fin:
data.append(json.loads(line))
return data
def create_doc_id2IPO_index(data):
id2data = defaultdict(list)
for d in data:
id2data[d["id"]].append(d["pred"]["label"])
id2result = {}
for doc_id, labels in id2data.items():
id2result[doc_id] = Counter(labels).most_common(1)[0][0]
return id2result
def create_field2class_index(et_fields_list):
field2class = defaultdict(set)
for ec in et_fields_list:
type_name = ec[0]
fields = ec[1]
for f in fields:
field2class[f].add(type_name)
return field2class
def reveal_instances_from_guessing(
pred_types: set,
mspans: List[dict],
field2class: DefaultDict[str, set],
guess_strategy: Optional[str] = "&",
):
event_list = []
if len(pred_types) <= 0:
# needs type guessing
guessed_types = set()
for pair in mspans:
if "Trigger" not in pair["mtype"]:
mspan = pair["mspan"]
mtype = pair["mtype"]
g_types = field2class[mtype]
if len(guessed_types) <= 0:
guessed_types.update(g_types)
else:
if guess_strategy == "&":
guessed_types.intersection_update(g_types)
elif guess_strategy == "|":
guessed_types.update(g_types)
else:
raise ValueError(
f"guess_strategy: {guess_strategy} not supported"
)
pred_types = guessed_types
# need finding
for pred_type in pred_types:
args = []
for pair in mspans:
if "Trigger" not in pair["mtype"]:
mspan = pair["mspan"]
mtype = pair["mtype"]
if mtype in event_type2event_class[pred_type].FIELDS:
args.append({"role": mtype, "argument": mspan})
if len(args) > 0:
event_list.append({"event_type": pred_type, "arguments": args})
return event_list
if __name__ == "__main__":
add_guessing_instance = False
guess_strategy = "&"
add_additional_arg = False
add_IPO_stage = False
# 合并可合并事件
merge_part_of_event = False
if add_IPO_stage:
# IPO_stage_result_filepath = "InferenceResults/test1-IPO_pred-large.json"
IPO_stage_result_filepath = "InferenceResults/test2_IPO_pred-large.json"
doc_id2IPO_stage_result = create_doc_id2IPO_index(
load_line_json(IPO_stage_result_filepath)
)
remove_comments = True
remove_triggers = True
"""test2"""
to_remove_filepath = "dueefin_PTPCG_woTgg_midImpt.json"
save_filepath = "ppsed_dueefin_PTPCG_woTgg_midImpt.json"
to_remove = load_line_json(to_remove_filepath)
field2class = create_field2class_index(event_type_fields_list)
mergable = 0
with open(save_filepath, "wt", encoding="utf-8") as fout:
for d in to_remove:
event_list = []
if len(d["event_list"]) <= 0:
if add_guessing_instance:
pred_types = set(d["comments"]["pred_types"])
event_list = reveal_instances_from_guessing(
pred_types,
d["comments"]["mspans"],
field2class,
guess_strategy=guess_strategy,
)
else:
pass
else:
for ins in d["event_list"]:
arguments = []
for arg_pair in ins["arguments"]:
if arg_pair["role"] == "Trigger" and remove_triggers:
continue
arguments.append(arg_pair)
if add_additional_arg:
exist_roles = {x["role"] for x in arguments}
for pair in d["comments"]["mspans"]:
tmp_arg = {"role": pair["mtype"], "argument": pair["mspan"]}
if (
pair["mtype"] not in exist_roles
and tmp_arg not in arguments
and pair["mtype"]
in event_type2event_class[ins["event_type"]].FIELDS
):
exist_roles.add(pair["mtype"])
arguments.append(tmp_arg)
ins["arguments"] = arguments
if (
add_IPO_stage
and ins["event_type"] == "公司上市"
and all(x["role"] != "环节" for x in ins["arguments"])
and d["id"] in doc_id2IPO_stage_result
):
ins["arguments"].append(
{"role": "环节", "argument": doc_id2IPO_stage_result[d["id"]]}
)
event_list.append(ins)
if merge_part_of_event:
event_list_merge_flag = [True for _ in range(len(event_list))]
for ins1, ins2 in combinations(enumerate(event_list), 2):
if ins1[1]["event_type"] == ins2[1]["event_type"]:
ins1_args = {
(arg["role"], arg["argument"])
for arg in ins1[1]["arguments"]
}
ins2_args = {
(arg["role"], arg["argument"])
for arg in ins2[1]["arguments"]
}
if ins1_args == ins2_args or ins2_args.issubset(ins1_args):
event_list_merge_flag[ins2[0]] = False
elif ins1_args.issubset(ins2_args):
event_list_merge_flag[ins1[0]] = False
new_event_list = []
for flag, events in zip(event_list_merge_flag, event_list):
if flag:
new_event_list.append(events)
else:
mergable += 1
else:
new_event_list = event_list
d["event_list"] = new_event_list
if remove_comments and "comments" in d:
del d["comments"]
fout.write(f"{json.dumps(d, ensure_ascii=False)}\n")
print("可合并的事件实例数：", mergable)