cyberbullying detection on social media texts
ABSTARCT : Cyberbullying is when someone is bullied using technology as an intermediary. Despite the fact that it has been a problem for many years, the impact on young people has just recently become more widely recognized.
Bullies thrive on social media platforms, and teens and children who use them are vulnerable to attacks. A copious amount of usergenerated communication data is generated by the widespread usage of social platforms netwroks by individuals.. Because of the prevalence of social public network, online bullying has become a severe issue in online web communication, and cyberbullying is getting aggrandized attention.
Cyberbullying has the potential to have a wide range of negative effects on an individual's life, including teen suicide. Twitter has provided a set of unique characteristics that have been included into the existing system, including activity, network, user, and tweet content.
Machine learning techniques recognize cyberbullying phrases existing in tweet content using these criteria. The identification of cyberbully words on Twitter is combined into one application in this proposed study, and cyberbully contents within tweet comments will be detected using Linear SVM.
This project comprises of a web-based application written in Django (a Python programming language), as well as a dashboard for users to track cyberbullying. An email is also sent to the recipients of this alert.
The prevalence of cybervictimisation depends on the conceptualisation used in describing cyberbullying, but also on research variables such as location and the number and age span of the participants. Nevertheless, the above studies demonstrate that online platforms are increasingly used for bullying, which is a cause for concern given its impact.
As shown by [9–11], cyberbullying may negatively impact the victim’s self-esteem, academic achievement and emotional well-being. [12] found that self-reported effects of cyberbullying include negative effects on school grades and feelings of sadness, anger, fear, and depression. In extreme cases, cyberbullying could even lead to self-harm and suicidal thoughts.
These findings demonstrate that cyberbullying is a serious problem the consequences of which can be dramatic. Early detection of cyberbullying attempts is therefore of key importance to youngsters’ mental well-being.
Successful detection depends on effective monitoring of online content, but the amount of information on the Web makes it practically unfeasible for moderators to monitor all user-generated content manually. To tackle this problem, intelligent systems are required that process this information in a fast way and automatically signal potential threats.
False Positives: Detection algorithms may misidentify non-abusive language as bullying, leading to unnecessary actions against users.
Context Sensitivity: The effectiveness of detection tools can vary greatly depending on the context in which language is used, making it challenging to accurately interpret intent.
Evasion Tactics: Cyberbullies may adapt their language to evade detection, using euphemisms, coded language, or emojis, which can undermine detection efforts.
Cultural Variability: Different cultures and communities may have varying definitions of what constitutes bullying, complicating the application of a one-size-fits-all detection system.
Data Privacy Concerns: Implementing detection tools may raise ethical issues related to user privacy and data handling, potentially leading to mistrust among users.
The system will begin by collecting data from various social media platforms through their APIs, ensuring a diverse and representative dataset. Preprocessing steps will include text cleaning, tokenization, and the extraction of key linguistic features such as sentiment scores and N-grams.
For detection, we will implement a hybrid approach combining traditional machine learning algorithms, such as Support Vector Machines (SVM) and Random Forests, with deep learning models like BERT and Long Short-Term Memory (LSTM) networks.
This dual strategy will enhance the model's ability to understand context and nuance, thereby reducing false positives and improving accuracy.
A user-friendly dashboard will be designed for real-time monitoring and reporting, allowing users to view flagged content and provide feedback on detection accuracy. This feedback loop will be integral for continuously refining the model, ensuring it adapts to evolving language trends and cultural contexts.
Early Intervention: Detection tools can identify harmful behavior early, allowing for timely intervention to support victims and prevent escalation.
Data-Driven Insights: Analyzing patterns in cyberbullying can provide valuable insights into its prevalence and impact, informing policy and educational efforts.
Increased Awareness: Implementing detection systems can raise awareness about cyberbullying, encouraging users to recognize and report abusive behavior.
Automated Monitoring: Automated tools can continuously monitor vast amounts of content, providing a level of oversight that would be impossible for human moderators alone.
Support for Victims: Effective detection systems can help connect victims with resources, support services, and reporting mechanisms, facilitating a safer online environment.
|