Spaces:
Sleeping
Sleeping
File size: 7,555 Bytes
a092d54 2c1f9dd a092d54 2c1f9dd a092d54 2c1f9dd a092d54 2c1f9dd a092d54 0ca5366 a092d54 0ca5366 a092d54 0ca5366 a092d54 0ca5366 a092d54 2c1f9dd a092d54 2c1f9dd a092d54 |
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 |
from __future__ import annotations
import bz2
from typing import TYPE_CHECKING, Literal
import pandas as pd
import spacy
from tqdm import tqdm
from app.constants import (
AMAZONREVIEWS_PATH,
AMAZONREVIEWS_URL,
IMDB50K_PATH,
IMDB50K_URL,
SENTIMENT140_PATH,
SENTIMENT140_URL,
TEST_DATASET_PATH,
TEST_DATASET_URL,
)
if TYPE_CHECKING:
from spacy.tokens import Doc
__all__ = ["load_data", "tokenize"]
try:
nlp = spacy.load("en_core_web_sm", disable=["tok2vec", "parser", "ner"])
except OSError:
print("Downloading spaCy model...")
from spacy.cli import download as spacy_download
spacy_download("en_core_web_sm")
nlp = spacy.load("en_core_web_sm", disable=["tok2vec", "parser", "ner"])
def _lemmatize(doc: Doc, threshold: int = 2) -> list[str]:
"""Lemmatize the provided text using spaCy.
Args:
doc: spaCy document
threshold: Minimum character length of tokens
Returns:
Lemmatized text
"""
return [
token.lemma_.lower().strip()
for token in doc
if not token.is_stop
and not token.is_punct
and not token.like_email
and not token.like_url
and not token.like_num
and not (len(token.lemma_) < threshold)
]
def tokenize(
text_data: list[str],
batch_size: int = 512,
n_jobs: int = 4,
character_threshold: int = 2,
show_progress: bool = True,
) -> list[list[str]]:
"""Tokenize the provided text using spaCy.
Args:
text_data: Text data to tokenize
batch_size: Batch size for tokenization
n_jobs: Number of parallel jobs
character_threshold: Minimum character length of tokens
show_progress: Whether to show a progress bar
Returns:
Tokenized text data
"""
return [
_lemmatize(doc, character_threshold)
for doc in tqdm(
nlp.pipe(text_data, batch_size=batch_size, n_process=n_jobs),
total=len(text_data),
disable=not show_progress,
)
]
def load_sentiment140(include_neutral: bool = False) -> tuple[list[str], list[int]]:
"""Load the sentiment140 dataset and make it suitable for use.
Args:
include_neutral: Whether to include neutral sentiment
Returns:
Text and label data
Raises:
FileNotFoundError: If the dataset is not found
"""
# Check if the dataset exists
if not SENTIMENT140_PATH.exists():
msg = (
f"Sentiment140 dataset not found at: '{SENTIMENT140_PATH}'\n"
"Please download the dataset from:\n"
f"{SENTIMENT140_URL}"
)
raise FileNotFoundError(msg)
# Load the dataset
data = pd.read_csv(
SENTIMENT140_PATH,
encoding="ISO-8859-1",
names=[
"target", # 0 = negative, 2 = neutral, 4 = positive
"id", # The id of the tweet
"date", # The date of the tweet
"flag", # The query, NO_QUERY if not present
"user", # The user that tweeted
"text", # The text of the tweet
],
)
# Ignore rows with neutral sentiment
if not include_neutral:
data = data[data["target"] != 2]
# Map sentiment values
data["sentiment"] = data["target"].map(
{
0: 0, # Negative
4: 1, # Positive
2: 2, # Neutral
},
)
# Return as lists
return data["text"].tolist(), data["sentiment"].tolist()
def load_amazonreviews(merge: bool = True) -> tuple[list[str], list[int]]:
"""Load the amazonreviews dataset and make it suitable for use.
Args:
merge: Whether to merge the test and train datasets (otherwise ignore test)
Returns:
Text and label data
Raises:
FileNotFoundError: If the dataset is not found
"""
# Check if the dataset exists
test_exists = AMAZONREVIEWS_PATH[0].exists() or not merge
train_exists = AMAZONREVIEWS_PATH[1].exists()
if not (test_exists and train_exists):
msg = (
f"Amazonreviews dataset not found at: '{AMAZONREVIEWS_PATH[0]}' and '{AMAZONREVIEWS_PATH[1]}'\n"
"Please download the dataset from:\n"
f"{AMAZONREVIEWS_URL}"
)
raise FileNotFoundError(msg)
# Load the datasets
dataset = []
with bz2.BZ2File(AMAZONREVIEWS_PATH[1]) as train_file:
dataset.extend([line.decode("utf-8") for line in train_file])
if merge:
with bz2.BZ2File(AMAZONREVIEWS_PATH[0]) as test_file:
dataset.extend([line.decode("utf-8") for line in test_file])
# Split the data into labels and text
labels, texts = zip(*(line.split(" ", 1) for line in dataset)) # NOTE: Occasionally OOM
# Map sentiment values
sentiments = [int(label.split("__label__")[1]) - 1 for label in labels]
# Return as lists
return texts, sentiments
def load_imdb50k() -> tuple[list[str], list[int]]:
"""Load the imdb50k dataset and make it suitable for use.
Returns:
Text and label data
Raises:
FileNotFoundError: If the dataset is not found
"""
# Check if the dataset exists
if not IMDB50K_PATH.exists():
msg = (
f"IMDB50K dataset not found at: '{IMDB50K_PATH}'\n"
"Please download the dataset from:\n"
f"{IMDB50K_URL}"
) # fmt: off
raise FileNotFoundError(msg)
# Load the dataset
data = pd.read_csv(IMDB50K_PATH)
# Map sentiment values
data["sentiment"] = data["sentiment"].map(
{
"positive": 1,
"negative": 0,
},
)
# Return as lists
return data["review"].tolist(), data["sentiment"].tolist()
def load_test(include_neutral: bool = False) -> tuple[list[str], list[int]]:
"""Load the test dataset and make it suitable for use.
Args:
include_neutral: Whether to include neutral sentiment
Returns:
Text and label data
Raises:
FileNotFoundError: If the dataset is not found
"""
# Check if the dataset exists
if not TEST_DATASET_PATH.exists():
msg = (
f"Test dataset not found at: '{TEST_DATASET_PATH}'\n"
"Please download the dataset from:\n"
f"{TEST_DATASET_URL}"
)
raise FileNotFoundError(msg)
# Load the dataset
data = pd.read_csv(TEST_DATASET_PATH)
# Ignore rows with neutral sentiment
if not include_neutral:
data = data[data["label"] != 1]
# Map sentiment values
data["label"] = data["label"].map(
{
0: 0, # Negative
1: 1, # Neutral
2: 2, # Positive
},
)
# Return as lists
return data["text"].tolist(), data["label"].tolist()
def load_data(dataset: Literal["sentiment140", "amazonreviews", "imdb50k", "test"]) -> tuple[list[str], list[int]]:
"""Load and preprocess the specified dataset.
Args:
dataset: Dataset to load
Returns:
Text and label data
Raises:
ValueError: If the dataset is not recognized
"""
match dataset:
case "sentiment140":
return load_sentiment140(include_neutral=False)
case "amazonreviews":
return load_amazonreviews(merge=True)
case "imdb50k":
return load_imdb50k()
case "test":
return load_test(include_neutral=False)
case _:
msg = f"Unknown dataset: {dataset}"
raise ValueError(msg)
|