Datasets:
Languages:
English
Multilinguality:
monolingual
Size Categories:
unknown
Language Creators:
machine-generated
Annotations Creators:
expert-generated
Source Datasets:
original
ArXiv:
Tags:
word-segmentation
License:
ruanchaves
commited on
Commit
•
f426013
1
Parent(s):
94d5c8a
Update stan_small.py
Browse files- stan_small.py +37 -11
stan_small.py
CHANGED
@@ -2,7 +2,6 @@
|
|
2 |
|
3 |
import datasets
|
4 |
import pandas as pd
|
5 |
-
import pickle5 as pickle
|
6 |
|
7 |
_CITATION = """
|
8 |
@misc{bansal2015deep,
|
@@ -55,22 +54,49 @@ class StanSmall(datasets.GeneratorBasedBuilder):
|
|
55 |
def _generate_examples(self, filepath):
|
56 |
|
57 |
def get_segmentation(row):
|
58 |
-
|
59 |
-
|
60 |
-
|
61 |
-
|
62 |
-
|
63 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
64 |
|
65 |
-
|
|
|
|
|
|
|
|
|
66 |
|
67 |
with open(filepath, 'rb') as f:
|
68 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
69 |
records = records.to_dict("records")
|
70 |
for idx, row in enumerate(records):
|
|
|
|
|
71 |
yield idx, {
|
72 |
"index": idx,
|
73 |
"hashtag": row["hashtags"],
|
74 |
-
"segmentation":
|
75 |
-
"alternatives":
|
76 |
}
|
2 |
|
3 |
import datasets
|
4 |
import pandas as pd
|
|
|
5 |
|
6 |
_CITATION = """
|
7 |
@misc{bansal2015deep,
|
54 |
def _generate_examples(self, filepath):
|
55 |
|
56 |
def get_segmentation(row):
|
57 |
+
needle = row["hashtags"]
|
58 |
+
haystack = row["goldtruths"][0].strip()
|
59 |
+
output = ""
|
60 |
+
iterator = iter(haystack)
|
61 |
+
for char in needle:
|
62 |
+
output += char
|
63 |
+
while True:
|
64 |
+
try:
|
65 |
+
next_char = next(iterator)
|
66 |
+
if next_char.lower() == char.lower():
|
67 |
+
break
|
68 |
+
elif next_char.isspace():
|
69 |
+
output = output[0:-1] + next_char + output[-1]
|
70 |
+
except StopIteration:
|
71 |
+
break
|
72 |
+
return output
|
73 |
|
74 |
+
def get_alternatives(row, segmentation):
|
75 |
+
alts = list(set([x.strip() for x in row["goldtruths"]]))
|
76 |
+
alts = [x for x in alts if x != segmentation]
|
77 |
+
alts = [{"segmentation": x} for x in alts]
|
78 |
+
return alts
|
79 |
|
80 |
with open(filepath, 'rb') as f:
|
81 |
+
try:
|
82 |
+
import pickle
|
83 |
+
records = pickle.load(f)
|
84 |
+
except ValueError:
|
85 |
+
try:
|
86 |
+
import pickle5 as pickle
|
87 |
+
records = pickle.load(f)
|
88 |
+
except ModuleNotFoundError:
|
89 |
+
raise ImportError(
|
90 |
+
"""To be able to use stan_small, you need to install the following dependencies['pickle5']
|
91 |
+
using 'pip install pickle5' for instance"""
|
92 |
+
)
|
93 |
records = records.to_dict("records")
|
94 |
for idx, row in enumerate(records):
|
95 |
+
segmentation = get_segmentation(row)
|
96 |
+
alternatives = get_alternatives(row, segmentation)
|
97 |
yield idx, {
|
98 |
"index": idx,
|
99 |
"hashtag": row["hashtags"],
|
100 |
+
"segmentation": segmentation,
|
101 |
+
"alternatives": alternatives
|
102 |
}
|