Cyberbullying detection in twitter space using lstm neural network

Abstract

Cyberbullying is a troubling and disconcerting online misconduct. It seems in different forms and mostly in textual formats on various social networking sites. Cyberbullying has become a common problem on online platforms, particularly social media platforms like Twitter. Detecting and contending cyberbullying is crucial to keeping users safe and healthy. Cyberbullying poses a serious risk to people’s mental health and entails effective detection and prevention. Various algorithms like Gated Recurrent Units (GRU), Recurrent Neural Networks (RNN), and Bidirectional Long Short Term Memory (BLSTM) are used to define the experimental results. This article focuses on developing a cyberbullying solution using Long Short Term Memory (LSTM), a deep learning technique for analyzing Twitter data and sequential data. This research emphasizesexploiting LSTMs' ability to capture data content and excerpt releva nt structures from the nature of Twitter data. The plan is comprising the collection of data, prioritization, LSTM model design, training, and evaluation. Experimental results prove that the theLSTMbased solution outperforms the basic method and demonstrates its accuracy and robustness in defining the nature of Twitter cyberbullying. The outcomes of this research help to meet the urgent need for cyberbullying detection technology and pave the way for future expansions in this field. Data pre-processing steps such as cleaning of text, tokenization, Stemming, Lemmatization, and removing stop words are used. After the pre-processing has been done, textual data that are already cleaned, have been passed to these algorithms of deep learning for forecasting purposes.process begins by collecting a large database of text tweets, including both cyberbullying and nonbullying ones. Using advanced techniques such as tokenization and rooting to convert the raw text into a format suitable for LSTM models. The proposed model was evaluated on a diverse and representative Twitter dataset, taking into account various criteria such as precision, recall, and F1 score. The results demonstrate the effectiveness of the LSTMbased method in identifying cyberbullying incidents, outperfor ming traditional machine learning, and demonstrating the ability to detect cyberbullying on Twitter.