Skip to content

A fast, robust Python library to check for offensive language in strings.

License

Notifications You must be signed in to change notification settings

Giovanni-Alzetta/profanity-check

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

28 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

profanity-check

Build Status release

A fast, robust Python library to check for profanity or offensive language in strings. Read more about how and why profanity-check was built in this blog post. You can also test out profanity-check in your browser.

How It Works

profanity-check uses a linear SVM model trained on 200k human-labeled samples of clean and profane text strings. Its model is simple but surprisingly effective, meaning profanity-check is both robust and extremely performant.

Why Use profanity-check?

No Explicit Blacklist

Many profanity detection libraries use a hard-coded list of bad words to detect and filter profanity. For example, profanity uses this wordlist, and even better-profanity still uses a wordlist. There are obviously glaring issues with this approach, and, while they might be performant, these libraries are not accurate at all.

A simple example for which profanity-check is better is the phrase "You cocksucker" - profanity thinks this is clean because it doesn't have "cocksucker" in its wordlist.

Performance

Other libraries like profanity-filter use more sophisticated methods that are much more accurate but at the cost of performance. A benchmark (performed December 2018 on a new 2018 Macbook Pro) using a Kaggle dataset of Wikipedia comments yielded roughly the following results:

Package 1 Prediction (ms) 10 Predictions (ms) 100 Predictions (ms)
profanity-check 0.2 0.5 3.5
profanity-filter 60 1200 13000
profanity 0.3 1.2 24

profanity-check is anywhere from 300 - 4000 times faster than profanity-filter in this benchmark!

Accuracy

This table speaks for itself:

Package Test Accuracy Balanced Test Accuracy Precision Recall F1 Score
profanity-check 95.0% 93.0% 86.1% 89.6% 0.88
profanity-filter 91.8% 83.6% 85.4% 70.2% 0.77
profanity 85.6% 65.1% 91.7% 30.8% 0.46

See the How section below for more details on the dataset used for these results.

Installation

$ pip install profanity-check

Usage

from profanity_check import predict, predict_prob

predict(['predict() takes an array and returns a 1 for each string if it is offensive, else 0.'])
# [0]

predict(['fuck you'])
# [1]

predict_prob(['predict_prob() takes an array and returns the probability each string is offensive'])
# [0.08686173]

predict_prob(['go to hell, you scum'])
# [0.7618861]

Note that both predict() and predict_prob return numpy arrays.

More on How/Why It Works

How

Special thanks to the authors of the datasets used in this project. profanity-check was trained on a combined dataset from 2 sources:

profanity-check relies heavily on the excellent scikit-learn library. It's mostly powered by scikit-learn classes CountVectorizer, LinearSVC, and CalibratedClassifierCV. It uses a Bag-of-words model to vectorize input strings before feeding them to a linear classifier.

Why

One simplified way you could think about why profanity-check works is this: during the training process, the model learns which words are "bad" and how "bad" they are because those words will appear more often in offensive texts. Thus, it's as if the training process is picking out the "bad" words out of all possible words and using those to make future predictions. This is better than just relying on arbitrary word blacklists chosen by humans!

Caveats

This library is far from perfect. For example, it has a hard time picking up on less common variants of swear words like "f4ck you" or "you b1tch" because they don't appear often enough in the training corpus. Never treat any prediction from this library as unquestionable truth, because it does and will make mistakes. Instead, use this library as a heuristic.

About

A fast, robust Python library to check for offensive language in strings.

Resources

License

Stars

Watchers

Forks

Packages

No packages published

Languages

  • Python 100.0%