Datasets:
1T Conte
commited on
Commit
•
e40c071
1
Parent(s):
b5c1167
first commit
Browse files- README.md +58 -0
- convert.py +58 -0
- reviews.parquet +3 -0
- reviews_polarity.parquet +3 -0
README.md
ADDED
@@ -0,0 +1,58 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
language:
|
3 |
+
- en
|
4 |
+
task_categories:
|
5 |
+
- text-classification
|
6 |
+
|
7 |
+
dataset_info:
|
8 |
+
- config_name: default
|
9 |
+
features:
|
10 |
+
- name: stars
|
11 |
+
dtype:
|
12 |
+
class_label:
|
13 |
+
names:
|
14 |
+
'0': 1 star
|
15 |
+
'1': 2 stars
|
16 |
+
'2': 3 stars
|
17 |
+
'3': 4 stars
|
18 |
+
'4': 5 stars
|
19 |
+
- name: date
|
20 |
+
dtype: string
|
21 |
+
- name: text
|
22 |
+
dtype: string
|
23 |
+
- name: useful
|
24 |
+
dtype: int16
|
25 |
+
- name: funny
|
26 |
+
dtype: int16
|
27 |
+
- name: cool
|
28 |
+
dtype: int16
|
29 |
+
- config_name: polarity
|
30 |
+
features:
|
31 |
+
- name: stars
|
32 |
+
dtype:
|
33 |
+
class_label:
|
34 |
+
names:
|
35 |
+
'0': negative
|
36 |
+
'1': positive
|
37 |
+
- name: date
|
38 |
+
dtype: string
|
39 |
+
- name: text
|
40 |
+
dtype: string
|
41 |
+
- name: useful
|
42 |
+
dtype: int16
|
43 |
+
- name: funny
|
44 |
+
dtype: int16
|
45 |
+
- name: cool
|
46 |
+
dtype: int16
|
47 |
+
|
48 |
+
configs:
|
49 |
+
- config_name: default
|
50 |
+
data_files:
|
51 |
+
- split: complete
|
52 |
+
path: "reviews.parquet"
|
53 |
+
default: true
|
54 |
+
- config_name: polarity
|
55 |
+
data_files:
|
56 |
+
- split: complete
|
57 |
+
path: "reviews_polarity.parquet"
|
58 |
+
---
|
convert.py
ADDED
@@ -0,0 +1,58 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import json
|
2 |
+
import random
|
3 |
+
|
4 |
+
import pandas as pd
|
5 |
+
|
6 |
+
|
7 |
+
def parse_review(line):
|
8 |
+
d = json.loads(line)
|
9 |
+
d.pop("review_id")
|
10 |
+
d.pop("user_id")
|
11 |
+
d.pop("business_id")
|
12 |
+
d["stars"] = int(d["stars"]) - 1 # from 0 to 4
|
13 |
+
|
14 |
+
return d
|
15 |
+
|
16 |
+
|
17 |
+
def remap2polarity(row):
|
18 |
+
if row["stars"] in [0, 1]:
|
19 |
+
row["stars"] = 0
|
20 |
+
elif row["stars"] in [3, 4]:
|
21 |
+
row["stars"] = 1
|
22 |
+
else:
|
23 |
+
raise ValueError("Invalid value")
|
24 |
+
|
25 |
+
return row
|
26 |
+
|
27 |
+
|
28 |
+
def run():
|
29 |
+
rows = []
|
30 |
+
|
31 |
+
with open(
|
32 |
+
"yelp_dataset/yelp_academic_dataset_review.json", "r", encoding="utf8"
|
33 |
+
) as file:
|
34 |
+
lines = file.readlines()
|
35 |
+
|
36 |
+
for line in lines:
|
37 |
+
obj = parse_review(line)
|
38 |
+
rows.append(obj)
|
39 |
+
|
40 |
+
df = pd.DataFrame(rows)
|
41 |
+
df.to_parquet("reviews.parquet")
|
42 |
+
|
43 |
+
# polarity
|
44 |
+
polarity_rows = [remap2polarity(row) for row in rows if row["stars"] != 2]
|
45 |
+
|
46 |
+
positive = [row for row in polarity_rows if row["stars"] == 1]
|
47 |
+
negative = [row for row in polarity_rows if row["stars"] == 0]
|
48 |
+
min_size = min(len(positive), len(negative))
|
49 |
+
|
50 |
+
polarity_rows = positive[:min_size] + negative[:min_size]
|
51 |
+
random.shuffle(polarity_rows)
|
52 |
+
|
53 |
+
df = pd.DataFrame(polarity_rows)
|
54 |
+
df.to_parquet("reviews_polarity.parquet")
|
55 |
+
|
56 |
+
|
57 |
+
if __name__ == "__main__":
|
58 |
+
run()
|
reviews.parquet
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:069f66f46b8b6def36fd346b9ca3c3ad3a81ae573d696865c8fcac47a01bc3fc
|
3 |
+
size 2636471449
|
reviews_polarity.parquet
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:c0db32db7029489aadb11e4d51cf8421727978e03e9a2c6f534e70657787d707
|
3 |
+
size 1301672500
|