Tweets Preprocessor

The tweets preprocessor module, developed by the AUTH team as part of the PlasticTwist Crowdsourcing module

Installation

The tweets-preprocessor module is not yet available trough PyPI, thus requiring manual import.

$ pip install -r requirements.txt

then execute utils/requirements_installer.py to install additional dependencies automatically.

Usage

The module was developed in a functional way and features a Fluent API. This allows the user to either call individual pre-processing methods or use the full_preprocess method to apply all of the pre-processing methods to his text.

The list of methods that can currently be used are:

• remove_urls - Removes all urls (e.g. 'https://ptwist.eu')
• remove_mentions - Removes all mentions (e.g. '@PlasticTwistBot')
• remove_hashtags - Removes all hashtags (e.g. '#plastictwist')
• remove_twitter_reserved_words - Removes Twitter reserved words (e.g. 'RT', 'via')
• remove_punctuation - Removes punctuation (e.g. '.', '!')
• remove_single_letter_words - Removes single-letter words (e.g. 'b', 'f')
• remove_blank_spaces - Removes blank spaces
• remove_stopwords - Removes stopwords (e.g. 'a', 'at', 'here')
  • has an extra_stopwords parameter (list) that allows users to add extra stopwords
• remove_profane_words - Removes profane words
• remove_numbers - Removes numbers (e.g. '2', '999')
  • has an preserve_years parameter (boolean) that allows users to choose whether or not years should be removed.

Examples

Using specific methods

from twitter_preprocessor import TwitterPreprocessor

p = TwitterPreprocessor('Some @ptwist text to be preprocessed. It contains 2 sentences. Best text 2019!')

p.remove_mentions().remove_punctuation().remove_numbers(preserve_years=True).remove_blank_spaces()
print(p.text)
# 'Some text to be preprocessed It contains sentences Best text 2019'

Full pre-processing

from twitter_preprocessor import TwitterPreprocessor

p = TwitterPreprocessor('RT @ptwist This text contains mentions, urls, some Twitter words and some stopwords to be preprocessed via https://example.com.')

p.fully_preprocess()
print(p.text)
# 'This text contains mentions urls Twitter words stopwords preprocessed'

License

This project is licensed under the GPL 3.0 license.

Credits

Developed and maintained by: vasisouv, alextsil, idimitriadis