Vulgar post detection using deep learning in facebook post (meta)
Sreekumar Nedumpally Raman, Narendran Sobanapuram Muruganandam, Vaishak S. Pai · A.A. Balkema eBooks · 2025
The rapid expansion of social media network has highlighted the necessity of efficient content moderation, especially with regard to spotting and eliminating offensive language. This article introduces an innovative method using bidirectional long short-term memory (Bi-LSTM) in recurrent neural net- works, to identify offensive content in social media messages. Our methodology targets textual information that may contain inflammatory language through a thorough data collection approach from various social media sources. Tokenization, lemmatization, and stop word removal are applied with care as preprocessing steps to ensure that the collected data is ready for deep learning analysis. The proposed model gives an accuracy of 51.8%. The findings of our deep learning experiment on vulgar post identification are encouraging and point to the possibility of using it to make online spaces safer. Because of the model&s;s capacity to recognize explicit language, content regulation may be enhanced, exposing users to far fewer harmful content and encouraging the development of more civil online communities. In order to prevent restricting genuine expression, the effectiveness of such a system depends on minimizing bias in the training data and making sure the model can handle contextual nuances.