-
Notifications
You must be signed in to change notification settings - Fork 2
/
Copy pathevaluate_submissions.py
168 lines (118 loc) · 5.89 KB
/
evaluate_submissions.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
# coding: utf8
import csv
from pathlib import Path
from score_test import evaluate_1, evaluate_2, evaluate_3
gold_folder = Path("test") / "gold"
def build_participants_table():
base = Path("submissions")
return [f for f in base.iterdir() if f.is_dir()]
def update(result, scenario, name):
for k,v in scenario.items():
result[name + '-' + k] = v
return result
def compute_taskA(d):
total_pre = d['correct_A'] + d['partial_A'] + d['spurious_A']
total_rec = d['correct_A'] + d['partial_A'] + d['missing_A']
pre = ((d['correct_A'] + 0.5 * d['partial_A']) / total_pre) if total_pre else 0
rec = ((d['correct_A'] + 0.5 * d['partial_A']) / total_rec) if total_rec else 0
d['task_A_pre'] = pre
d['task_A_rec'] = rec
d['task_A_f1'] = ( 2 * pre * rec ) / ( pre + rec ) if ( pre + rec > 0 ) else 0
return d
def compute_taskB(d):
total_acc = d['correct_B'] + d['incorrect_B']
acc = ( d['correct_B'] / total_acc) if total_acc else 0
d['task_B_acc'] = acc
return d
def compute_taskC(d):
total_pre = d['correct_C'] + d['spurious_C']
total_rec = d['correct_C'] + d['missing_C']
pre = ((d['correct_C']) / total_pre) if total_pre else 0
rec = ((d['correct_C']) / total_rec) if total_rec else 0
d['task_C_pre'] = pre
d['task_C_rec'] = rec
d['task_C_f1'] = ( 2 * pre * rec ) / ( pre + rec ) if ( pre + rec > 0 ) else 0
return d
def compute_scenario1(d):
total_pre = d['correct_A'] + d['partial_A'] + d['spurious_A'] + d['correct_B'] + d['incorrect_B'] + d['correct_C'] + d['spurious_C']
total_rec = d['correct_A'] + d['partial_A'] + d['missing_A'] + d['correct_B'] + d['incorrect_B'] + d['correct_C'] + d['missing_C']
pre = ( d['correct_A'] + 0.5 * d['partial_A'] + d['correct_B'] + d['correct_C'] ) / total_pre if total_pre > 0 else 0
rec = ( d['correct_A'] + 0.5 * d['partial_A'] + d['correct_B'] + d['correct_C'] ) / total_rec if total_rec > 0 else 0
d['total_micro_pre'] = pre
d['total_micro_rec'] = rec
d['total_micro_f1'] = ( 2 * pre * rec ) / ( pre + rec ) if ( pre + rec > 0 ) else 0
def compute_scenario2(d):
total_pre = d['correct_B'] + d['incorrect_B'] + d['correct_C'] + d['spurious_C']
total_rec = d['correct_B'] + d['incorrect_B'] + d['correct_C'] + d['missing_C']
pre = ( d['correct_B'] + d['correct_C'] ) / total_pre if total_pre > 0 else 0
rec = ( d['correct_B'] + d['correct_C'] ) / total_rec if total_rec > 0 else 0
d['total_micro_pre'] = pre
d['total_micro_rec'] = rec
d['total_micro_f1'] = ( 2 * pre * rec ) / ( pre + rec ) if ( pre + rec > 0 ) else 0
def compute_scenario3(d):
total_pre = d['correct_C'] + d['spurious_C']
total_rec = d['correct_C'] + d['missing_C']
pre = ( d['correct_C'] ) / total_pre if total_pre > 0 else 0
rec = ( d['correct_C'] ) / total_rec if total_rec > 0 else 0
d['total_micro_pre'] = pre
d['total_micro_rec'] = rec
d['total_micro_f1'] = ( 2 * pre * rec ) / ( pre + rec ) if ( pre + rec > 0 ) else 0
def compute_micro_f1(d1, d2, d3):
total_pre_1 = d1['correct_A'] + d1['partial_A'] + d1['spurious_A'] + d1['correct_B'] + d1['incorrect_B'] + d1['correct_C'] + d1['spurious_C']
total_rec_1 = d1['correct_A'] + d1['partial_A'] + d1['missing_A'] + d1['correct_B'] + d1['incorrect_B'] + d1['correct_C'] + d1['missing_C']
correct_1 = d1['correct_A'] + 0.5 * d1['partial_A'] + d1['correct_B'] + d1['correct_C']
total_pre_2 = d2['correct_B'] + d2['incorrect_B'] + d2['correct_C'] + d2['spurious_C']
total_rec_2 = d2['correct_B'] + d2['incorrect_B'] + d2['correct_C'] + d2['missing_C']
correct_2 = d2['correct_B'] + d2['correct_C']
total_pre_3 = d3['correct_C'] + d3['spurious_C']
total_rec_3 = d3['correct_C'] + d3['missing_C']
correct_3 = d3['correct_C']
total_pre = total_pre_1 + total_pre_2 + total_pre_3
total_rec = total_rec_1 + total_rec_2 + total_rec_3
correct = correct_1 + correct_2 + correct_3
pre = correct / total_pre if total_pre > 0 else 0
rec = correct / total_rec if total_rec > 0 else 0
return 2 * pre * rec / ( pre + rec ) if ( pre + rec ) > 0 else 0
def evaluate_participant(submit_folder):
scenario1 = evaluate_1("scenario1.txt", gold_folder / "scenario1-ABC", submit_folder / "scenario1-ABC")
scenario2 = evaluate_2("scenario2.txt", gold_folder / "scenario2-BC", submit_folder / "scenario2-BC")
scenario3 = evaluate_3("scenario3.txt", gold_folder / "scenario3-C", submit_folder / "scenario3-C")
compute_taskA(scenario1)
compute_taskB(scenario1)
compute_taskC(scenario1)
compute_scenario1(scenario1)
compute_taskB(scenario2)
compute_taskC(scenario2)
compute_scenario2(scenario2)
compute_taskC(scenario3)
compute_scenario3(scenario3)
result = dict()
update(result, scenario1, 'S1')
update(result, scenario2, 'S2')
update(result, scenario3, 'S3')
result['Name'] = submit_folder.name.split('-')[0].strip()
result['Z-average_f1'] = ( scenario1['total_micro_f1'] + scenario2['total_micro_f1'] + scenario3['total_micro_f1'] ) / 3
result['Z-average_tasks'] = ( scenario1['task_A_f1'] + scenario2['task_B_acc'] + scenario3['task_C_f1'] ) / 3
result['Z-micro_f1'] = compute_micro_f1(scenario1, scenario2, scenario3)
return result
def evaluate_all():
results = []
for folder in build_participants_table():
results.append(evaluate_participant(folder))
keys = list(results[0].keys())
keys.sort()
with open('results.tsv', 'w') as fp:
for key in keys:
fp.write(key + "\t")
for item in results:
v = item[key]
if isinstance(v, str):
fp.write(v)
elif isinstance(v, int):
fp.write(str(v))
elif isinstance(v, float):
fp.write("%.3f" % v)
fp.write("\t")
fp.write("\n")
if __name__ == '__main__':
evaluate_all()