LeetSpeaker App using pyleetspeak package. pyleetspeak is an innovative Python package designed to simulate multilingual content evasion through word camouflage techniques such as leetspeak, punctuation insertion, and syllable inversion. Developed to combat content moderation evasion on social networks, LeetSpeaker supports over 20 languages, making it a versatile tool for researchers and developers. The package also functions as a data generator, using KeyBERT for semantically relevant word extraction and Spacy for data annotation. Open-source and customizable, LeetSpeaker aids in enhancing online security and serves as a data augmentation tool for AI systems.
This tools exemplifies how NLP models can be trained to recognize camouflaged entities in both monolingual and multilingual contexts. The tool unique aspect is that it focuses on word camouflage techniques, making it a valuable resource for research on content moderation evasion. The dataset is part of a larger effort to combat information disorders on social networks, and it's publicly available for research and application, promoting transparency and reproducibility in the field. The dataset is a part of a paper publication.
If you find it usefull please cite these contributions:
@article{HUERTASGARCIA2023110552,
title = {Countering malicious content moderation evasion in online social networks: Simulation and detection of word camouflage},
journal = {Applied Soft Computing},
volume = {145},
pages = {110552},
year = {2023},
issn = {1568-4946},
doi = {https://doi.org/10.1016/j.asoc.2023.110552},
author = {Álvaro Huertas-García and Alejandro Martín and Javier Huertas-Tato and David Camacho},
}