ruanchaves
commited on
Commit
·
d3df06b
1
Parent(s):
36bcc6b
Update stan_large.py
Browse files- stan_large.py +38 -11
stan_large.py
CHANGED
@@ -1,8 +1,8 @@
|
|
1 |
"""STAN large dataset"""
|
2 |
|
|
|
3 |
import datasets
|
4 |
import pandas as pd
|
5 |
-
import pickle5 as pickle
|
6 |
|
7 |
_CITATION = """
|
8 |
@inproceedings{maddela-etal-2019-multi,
|
@@ -77,22 +77,49 @@ class StanLarge(datasets.GeneratorBasedBuilder):
|
|
77 |
def _generate_examples(self, filepath):
|
78 |
|
79 |
def get_segmentation(row):
|
80 |
-
|
81 |
-
|
82 |
-
|
83 |
-
|
84 |
-
|
85 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
86 |
|
87 |
-
|
|
|
|
|
|
|
|
|
88 |
|
89 |
with open(filepath, 'rb') as f:
|
90 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
91 |
records = records.to_dict("records")
|
92 |
for idx, row in enumerate(records):
|
|
|
|
|
93 |
yield idx, {
|
94 |
"index": idx,
|
95 |
"hashtag": row["hashtags"],
|
96 |
-
"segmentation":
|
97 |
-
"alternatives":
|
98 |
}
|
|
|
1 |
"""STAN large dataset"""
|
2 |
|
3 |
+
from multiprocessing.sharedctypes import Value
|
4 |
import datasets
|
5 |
import pandas as pd
|
|
|
6 |
|
7 |
_CITATION = """
|
8 |
@inproceedings{maddela-etal-2019-multi,
|
|
|
77 |
def _generate_examples(self, filepath):
|
78 |
|
79 |
def get_segmentation(row):
|
80 |
+
needle = row["hashtags"]
|
81 |
+
haystack = row["goldtruths"][0].strip()
|
82 |
+
output = ""
|
83 |
+
iterator = iter(haystack)
|
84 |
+
for char in needle:
|
85 |
+
output += char
|
86 |
+
while True:
|
87 |
+
try:
|
88 |
+
next_char = next(iterator)
|
89 |
+
if next_char.lower() == char.lower():
|
90 |
+
break
|
91 |
+
elif next_char.isspace():
|
92 |
+
output = output[0:-1] + next_char + output[-1]
|
93 |
+
except StopIteration:
|
94 |
+
break
|
95 |
+
return output
|
96 |
|
97 |
+
def get_alternatives(row, segmentation):
|
98 |
+
alts = list(set([x.strip() for x in row["goldtruths"]]))
|
99 |
+
alts = [x for x in alts if x != segmentation]
|
100 |
+
alts = [{"segmentation": x} for x in alts]
|
101 |
+
return alts
|
102 |
|
103 |
with open(filepath, 'rb') as f:
|
104 |
+
try:
|
105 |
+
import pickle
|
106 |
+
records = pickle.load(f)
|
107 |
+
except ValueError:
|
108 |
+
try:
|
109 |
+
import pickle5 as pickle
|
110 |
+
records = pickle.load(f)
|
111 |
+
except ModuleNotFoundError:
|
112 |
+
raise ImportError(
|
113 |
+
"""To be able to use stan_large, you need to install the following dependencies['pickle5']
|
114 |
+
using 'pip install pickle5' for instance"""
|
115 |
+
)
|
116 |
records = records.to_dict("records")
|
117 |
for idx, row in enumerate(records):
|
118 |
+
segmentation = get_segmentation(row)
|
119 |
+
alternatives = get_alternatives(row, segmentation)
|
120 |
yield idx, {
|
121 |
"index": idx,
|
122 |
"hashtag": row["hashtags"],
|
123 |
+
"segmentation": segmentation,
|
124 |
+
"alternatives": alternatives
|
125 |
}
|