A basic and simple yet powerful Python library to detect toxicity/profanity of a review or list of reveiws.
Project description
cuss_inspect
How It Works
cuss_inspect
is a logistic regression based model trained on 180K+ reviews and tested on 24K+ reviews.
Performance
1 Prediction (ms) | 10 Predictions (ms) | 100 Predictions (ms) | 1000 Predictions (ms) | 10000 Predictions (ms) | |
---|---|---|---|---|---|
cuss_inspect | 0.2 | 0.3 | 0.8 | 4.3 | 24.7 |
Accuracy
Precision | Recall | F1 Score | |
---|---|---|---|
0 | 0.84 | 0.94 | 0.89 |
1 | 0.99 | 0.96 | 0.98 |
Accuracy | 0.96 | ||
macro avg | 0.91 | 0.95 | 0.93 |
weighted avg | 0.96 | 0.96 | 0.96 |
Receiver Operating Characteristics
Installation
$ pip install cuss_inspect
Usage
from cuss_inspect import predict, predict_prob
# for simple string
text_0 = "this is simple review. you have done a good job"
print(predict(text_0))
# [0]
print(predict_prob(text_0)
# [0.05]
text_1 = "son of a bitch"
print(predict(text_1))
# [1]
print(predict_prob(text_1)
# [1.]
# for list of inputs
test = ['who are you?' , 'what do you want?' , 'son of a dog' , 'how the hell can you say that' , 'fuck it']
print(predict(test))
# [0 0 1 1 1]
print(predict_prob(test))
# [0.12 0.22 0.55 0.96 1.]
*predict()
and predict_prob
return numpy
arrays.
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
cuss_inspect-1.0.2b0.tar.gz
(2.7 kB
view hashes)
Built Distribution
Close
Hashes for cuss_inspect-1.0.2b0-py3-none-any.whl
Algorithm | Hash digest | |
---|---|---|
SHA256 | 5adef3094820df66882b35bc2be6a74e6cd00a6b8e226d9658ee79f67c51e1a3 |
|
MD5 | bcd73e4d5c9108060e3cb9a892a185c0 |
|
BLAKE2b-256 | 5a22a29b8bfa0f5d649e663abcb020ffd4d1c062e9af32e8bff4dcb9d298e59a |