Cyberbullying
ABSTRACT:
Social networking sites become part of everyday life. Because of the increase and fame in social networking sites, the appearance of persons is online is common now. People regularly consult their secret or public talk online over social network sites. People are in touch with their siblings or friends via social networking sites or applications. So, Social network gives a platform to communicate with their known become easier. Krishna in 2015 represented that social networking platform reduces the transmission and response time. Today various types of social networking sites and apps are present online e.g. Facebook, Twitter, and Whatsapp, etc. [4]. With the improvement of technology, the craze of social networking platforms is increasing. People now share their information easily using computers, mobile phones, etc. However, this has lead to the growth of cyber-criminal acts, for example, cyberbullying which has become a worldwide epidemic. Cyberbullying is the use of electronic communication to bully a person by sending harmful messages using social media, instant messaging, or through digital communications. It has developed out as a platform for insulting, degrading a person which can affect the person either physically or emotionally and sometimes leading to suicidal attempts in the worst case. The main issue in checking to cyberbully is detecting its occurrence so that appropriate action can be taken at the initial stages. To overcome this problem, many methods and techniques had been worked upon until now to control this problem. This paper is a survey covering cyberbullying and cyberbullying detection techniques.
Keywords: cyberbully detection, facebook, cybercrime, twitter, safe chat
INTRODUCTION
Cyberbullying is a type of bullying that demands place using computerized technology including electronic devices and equipment such as cell phones, computers through social media, text messages, chats, etc [1]. Examples of cyberbullying include mean text messages, rumors that can be very embarrassing to the concerned person. This can occur at any time and results online and the text messages and images can be posted anonymously which can be spread quickly to a very large audience. Modern-day technology is a boon and cannot be accused of cyberbullying. Social networking sites are used for positive activities, like connecting kids with friends and family, helping pupils with school, and for fun. But these tools can also be used to hurt other people. Whether made online or offline the results of bullying are similar. Cyberbullying is also defined as “deliberate and repeated harm inflicted through the medium of electronic text [2]”.It mainly targets children and adolescents as they are most active on social networks. With Web 2.0 providing easy and ubiquitous online access, cybersecurity is becoming an important concern. Some of the most common forms of cyberbullying are as follows [3].
Flaming: online discussions and conflicts using obscene and abusive language.
Harassment: Frequently sending cruel, offensive, or bullying messages.
Denigration: Exposing the secrets of a person or gossips to damage the reputation of a person.
Impersonation: Breaking into the victim’s account and sending emails.
Trickery: Cheating the victim into disclosing sore information and giving it to others.
Interactive Gaming: Most gaming consoles enable people to unite and play online giving a chance to hurt using chats and comments.
With the exponential increase of social media users, cyberbullying has appeared as a form of bullying through electronic communications. Social networks provide a rich environment for bullies to uses these networks as vulnerable to attacks against victims. Given the consequences of cyberbullying on victims, it is necessary to find suitable actions to detect and prevent it. Cyber Bullying can be described as the use of data and information technology by a person or a group of users to irritate other users. This is the national health problem nowadays. Earlier harassing has been done by the people who are in contact or to whom physically meet regularly. But nowadays with the laptop or cell phone person can bully their victim irrespective of the geographical area via the internet. Sometimes, it is very difficult to detect a person who is behind the whole scenario [12]. The problem of cyberbullying is very common in the youngster. By advanced technology bullying transferred from the physical to virtual era. So cyberbullying refers to any behavior achieved by electronic media with somebody or a group of peoples to message someone to hurt or distress.
RELATED WORK
Detection of cyberbullying and the requirement of subsequent precautionary measures are the main programs of action to combat cyberbullying. The proposed method is an efficient method to detect cyberbullying activities on social media. The detection method can identify the presence of cyberbullying terms and classify cyberbullying exercises in a social network such as Harassment, Flaming, Terrorism, and Racism using Fuzzy logic and Genetic algorithm. The effectiveness of the system is increased using the Fuzzy rule set to retrieve relevant data for classification from the input. In the proposed method Genetic algorithm is also used, for optimizing the parameters and obtaining precise output. Using online available applications and tools. eBlaster, Net Nanny, cloud9, and IamBigBrother are some available commercial tools available to protect children from Internet predators. They work based on Packet sniffing. These tools scan all the outgoing and incoming traffic in a network and then apply a filter. But the difficulty with these tools is that they are based on a simple keyword matching and therefore its exactness is questioned. To overcome this limit, SafeChat was introduced. This software used the WinpCap library used by Wireshark. SafeChat supported Open System for Communication in Realtime (OSCAR) protocol. But the documentation of this protocol failed as it was not compatible with other protocols. Later SafeChat 2.0 was released as the new version of SafeChat. It is a third-party plug-in for the Pidgin, an open-source instant messaging system. It does predator detection algorithms to classify chat members as potential predators. [5]
Singh in 2019 investigates feasible combinations of various preprocessing, feature adoption, and review methodologies using the cuckoo search metaheuristic approach. That method seeks to improve the review of a content-based cybercrime detection system. For this examination, four publicly free datasets for cyberbullying detection have been used for estimating the effectiveness of the suggested algorithm. The algorithm compared with three novel cyberbullying detection models based on various evaluation parameters. These parameters included precision, recall, and f-measure. The results show the effectiveness of the suggested system. This system beat other recent techniques on all the datasets, providing high predictive recall value via tenfold cross-validation.[6]
Dadvar et. al examine the conclusions of recent research in this regard. He successfully represented the findings of this literature and validated their findings using the same datasets, namely Twitter, Wikipedia, and Formspring, used by the authors. Then he expanded their work by using the generated methods on a new YouTube dataset (~54k posts by ~4k users) and examined the appearance of the models in new social media programs. He assessed the performance of the models trained on one platform to another platform. Later that, he shows that the deep learning-based models were better than machine learning models previously applied to the same YouTube dataset. He also believes that the deep learning-based models can also benefit from mixing other origins of information and studying into the impact of profile knowledge of the users in social networks.[7]
- Zois et. all formulate cyberbullying detection as a sequential hypothesis testing problem. Based on this formulation, they suggest a novel algorithm designed to reduce the time to construct a cyber-bullying alert by drastically reducing the number of feature evaluations necessary for a decision to be made. They prove the effectiveness of their approach using a real-world dataset from Twitter, one of the top five networks with the highest percentage of users reporting cyberbullying instances. They show that their approach is highly scalable while not reducing accuracy for scalability.[8]
- Yao, et al propose a novel algorithm to drastically decrease the number of features used in classification. They prove the utility, scalability, and responsiveness of their approach using a real-world dataset from Instagram, the online social media platform with the highest percentage of users reporting undergoing cyberbullying. Their approach improves recall by a staggering 700%, while at the same time decreasing the average number of points by up to 99.82% related to state-of-the-art supervised cyberbullying detection methods, learning methods that require weak supervision, and traditional offline feature selection and dimensionality reducing techniques.[9]
Krishna B. Kansara, Narendra M. Shekokar proposed a framework in 2015 for identifying negative online interactions in terms of offensive contents carried out through text messages as well as images. The sequence of text & image analysis techniques is considered as a suitable platform for the detection of potential cyberbullying warnings.[4]. Mounir et al. 2019 Machine learning can be helpful to detect language models of the bullies and hence can generate a model to automatically detect cyberbullying activities. They use a supervised machine learning approach for detecting and preventing cyberbullying. Several classifiers are used to train and identify bullying actions. The evaluation of the proposed approach on cyberbullying dataset shows that the Neural Network works better and achieves the accuracy of 92.8% and SVM achieves 90.3. Also, NN beats other classifiers of related work on the same dataset.[10]. Al-gardadi et. al have proposed a set of unique features derived from Twitter; network, activity, user, and tweet content, based on these comments, they developed a supervised machine learning solution for identifying cyberbullying on Twitter. Model-based on their proposed features achieved results with an area under the receiver-operating characteristic curve of 0.943 and an f-measure of 0.936. These results indicate that the suggested model based on certain features gives a feasible solution to detecting Cyberbullying in online communication environments. Finally, they match results obtained using our proposed features with the result achieved from two baseline features. The comparison outcomes show the significance of the suggested features.[11].
CONCLUSION
With the vast use of the Internet, people communicate with other people. Which leads to the chance of misusing new technology? These methods lead to cyberbullying. Our literature review illustrates the cyberbullying detection techniques adopted so far to address this problem. We proposed machine learning techniques to detect cyberbullying. These methods make automatic detection of cyberbullying information in social media text and this could help to construct a healthy and safe social media environment for the people. So these methods can be worked upon by getting new datasets and then applying various machine learning algorithms on them to obtained desired accuracy. Sentiment analysis can also be performed on the social media data by extracting data from various social networking sites using the available datasets and tools to recognize the appearance or absence of cyberbullying by determining the text at sentence level or document level. Weka, Rapidminer, R, Orange are some of the tools that can be used for this purpose.
REFERENCES
1] https://www.stopbullying.gov/cyberbullying/
[2] J.Patchin, & S. Hinduja, “Bullies move beyond the schoolyard; a preliminary look at cyberbullying.” Youth violence and juvenile justice.4:2 (2006). 148-169.
[3] Sourabh Parime, Vaibhav Suri “Cyberbullying Detection and Prevention: Data Mining and Psychological Perspective”, 2014 International Conference on Circuit, Power and Computing Technologies [ICCPCT]
[4]. Krishna B. Kansara and Narendra M. Shekokar “A framework for cyberbullying Detection in Social Network” INPRESSCO: Feb 2015 Vol.5 No.1 E-ISSN 2277-4106
[5]. Shrivastava, Sonika & Pateriya, R.K.. (2017). A Review of Cyberbullying Detection in Social Networking.
[6]. Singh, A., Kaur, M. Detection Framework for Content-Based Cybercrime in Online Social Networks Using Metaheuristic Approach. Arab J Sci Eng 45, 2705–2719 (2020). https://doi.org/10.1007/s13369-019-04125-w
[7]. Dadvar, Maral & Eckert, Kai. (2018). Cyberbullying Detection in Social Networks Using Deep Learning Based Models; A Reproducibility Study. 10.13140/RG.2.2.16187.87846.
[8]. D. Zois, A. Kapodistria, M. Yao and C. Chelmis, “Optimal Online Cyberbullying Detection,” 2018 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), Calgary, AB, 2018, pp. 2017-2021, doi: 10.1109/ICASSP.2018.8462092.
[9]. M. Yao, C. Chelmis and D. Zois, “Cyberbullying Detection on Instagram with Optimal Online Feature Selection,” 2018 IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining (ASONAM), Barcelona, 2018, pp. 401-408, doi: 10.1109/ASONAM.2018.8508329.
[10]. Mounir, John & Nashaat, Mohamed & Ahmed, Mostafaa & Emad, Zeyad & Amer, Eslam & Mohammed, Ammar. (2019). Social Media Cyberbullying Detection using Machine Learning. International Journal of Advanced Computer Science and Applications. 10. 703-707. 10.14569/IJACSA.2019.0100587.
[11]. Al-Garadi, M. A., Varathan, K. D., & Ravana, S. D. (2016). Cybercrime detection in online communications: the experimental case of cyberbullying detection in the Twitter network. Computers in Human Behavior, 63, 433-443. doi:10.1016/j.chb.2016.05.051
[12]. Belsey, B. (2004).Cyberbullying. Retrieved from www.cyberbullying.ca