Aggressive Text Detection for Cyberbullying

Bosque, Laura P. Del; Garza, Sara E.

doi:10.1007/978-3-319-13647-9_21

Cited by 31 publications

(28 citation statements)

References 7 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Over the last decade, the body of literature on automated detection of cyberbullying has been growing, especially on the topic of detecting cyberbullying from social media networks like Twitter [7]- [13], Instagram [10], [11], [14]- [16] and YouTube [17]- [19]. This body of research has been working towards automated cyberbullying detection using either rulebased models [12], [19], [20], conventional machine learning models [16], [18], [19], [21], or deep learning models [13], [21]- [23].…”

Section: Introductionmentioning

confidence: 99%

When the Timeline Meets the Pipeline: A Survey on Automated Cyberbullying Detection

et al. 2021

View full text Add to dashboard Cite

Web 2.0 helped user-generated platforms to spread widely. Unfortunately, it also allowed for cyberbullying to spread. Cyberbullying has negative effects that could lead to cases of depression and low self-esteem. It has become crucial to develop tools for automated cyberbullying detection. The research on developing these tools has been growing over the last decade, especially with the recent advances in machine learning and natural language processing. Given the large body of work on this topic, it is vital to critically review the literature on cyberbullying within the context of these latest advances. In this paper, we conduct a survey on the automated detection of cyberbullying. Our survey sheds light on some challenges and limitations for the field. The challenges range from defining cyberbullying, data collection and feature representation to model selection, training and evaluation. We also provide some suggestions for improving the task of cyberbullying detection. In addition to the survey, we propose to improve the task of cyberbullying detection by addressing some of the raised limitations: 1) Using recent contextual language models like BERT for the detection of cyberbullying; 2) Using slang-based word embeddings to generate better representations of the cyberbullying-related datasets. Our results show that BERT outperforms state-of-the-art cyberbullying detection models and deep learning models. The results also show that deep learning models initialised with slang-based word embeddings outperform deep learning models initialised with traditional word embeddings.

show abstract

Section: Introductionmentioning

confidence: 99%

When the Timeline Meets the Pipeline: A Survey on Automated Cyberbullying Detection

et al. 2021

View full text Add to dashboard Cite

show abstract

“…To a lesser extent, researchers have analyzed the content of tweets such as sentiment (Choi et al, 2014; Resnik, Bellmore, Xu, & Zhu, 2016), hashtags (Calvin et al, 2015), and the role of the audience or online bystanders (Cocea, 2016). Previous Twitter research suggests that each approach to text analysis includes both strengths and weakness, for example, machine learning faces challenges when coding slang and “informal text” (Del Bosque & Garza, 2014), while human analysis might include personal bias due to individual experience. Recognizing this, the current study included both Linguistic Inquiry and Word Count (LIWC) software–generated analysis as well as human coding and analysis.…”

mentioning

confidence: 99%

OMG U #Cyberbully! An Exploration of Public Discourse About Cyberbullying on Twitter

2018

View full text Add to dashboard Cite

Cyberbullying, defined as bullying that takes place using technology, includes similar tactics found in traditional bullying as well as unique approaches such as viral repetition. Nationally, prevalence rates for cyberbullying range from 10% to as high as 40% of school-aged children, depending on the definition and measurement tool applied. The current study examines public tweets with keywords and hashtags related to cyberbullying posted during May 2016, using both human evaluation and computer examination to answer the following research questions: (1) What is the sentiment of tweets using cyberbullying keywords/hashtags? (2) What is the thematic content of the tweets? (3) What is the relationship between coding by researchers versus automated coding by Linguistic Inquiry and Word Count (LIWC) software? and (4) What is the content of the URLs attached to the tweets? A unique aspect of this study is the examination of the content of URLs included in the tweets, with the finding that the majority of the accessible URL references were to material that was positively focused. The majority of sample tweets referred to a cyberbully situation, contributed to a negative atmosphere, included references to known individuals, and suggested ongoing cyberbullying events. Results from this study suggest an opportunity for researchers, educators, and public health practitioners to use discourse on social media to inform interventions, to educate and share information, and to promote social well-being and mental health.

show abstract

“…The detection of aggressive text messages was performed using the approach proposed by Del Bosque and Garza [9], which concerns an unsupervised lexiconbased, term-counting strategy that identifies profane words. In summary, this approach, given a set M of messages, assigns a score sc i to each message m i .…”

Section: Aggressive Text Message Detectionmentioning

confidence: 99%

“…In that sense, a user that sends messages with a particular frequency and aggressiveness score is considered as an alleged aggressor or bully, while a user that receives messages with a particular frequency and aggressiveness score is considered as an alleged victim. For this case, that particular frequency is two or more messages within M (considering repetition) and that particular aggressiveness score is sc i ≥ 5 (considering this is the middle point of the scale used by Del Bosque and Garza [9]).…”

Section: Alleged Aggressor and Victim Detectionmentioning

confidence: 99%

“…A message dataset was collected from Twitter using the API and the methodology described in the works by Del Bosque and Garza [9] and Escalante et al [13], which in summary consists of obtaining directed messages with seed words (aggressive words, usually) and manually annotating these messages. An extensive list of aggressive words (shown in Table 1) was used to collect messages; these words were gathered with the aid of a native English speaker.…”

Section: Setupmentioning

confidence: 99%

See 1 more Smart Citation

Cyberbullying Detection in Social Networks: A Multi-Stage Approach

Bosque¹,

Villareal²

2019

RCS

Self Cite

View full text Add to dashboard Cite

Cyberbullying, defined as the violent harassment of an individual towards a victim in electronic media, is a serious problem nowadays. If not prevented or mitigated, it can lead to affective disorders, poor academic performance, problems in social relationships, andultimately-to suicide attempts in youngsters and children. Because manual supervision in social networks (a space where cyberbullying can naturally occur) is laborious, automated approaches for cyberbullying detection are desirable. However, a considerable number of approaches treat this problem as merely aggressive text message detection, without considering the frequency of the harassment. The approach proposed in this work, in contrast, views cyberbullying detection as a process of three consecutive stages: aggressive text message detection, alleged aggressor and victim detection, and cyberbullying case detection. This approach was tested using a dataset extracted from Twitter; the approach obtained an F-score of 0.947 for the positive (cyberbullying) case.

show abstract

Aggressive Text Detection for Cyberbullying

Cited by 31 publications

References 7 publications

When the Timeline Meets the Pipeline: A Survey on Automated Cyberbullying Detection

When the Timeline Meets the Pipeline: A Survey on Automated Cyberbullying Detection

OMG U #Cyberbully! An Exploration of Public Discourse About Cyberbullying on Twitter

Cyberbullying Detection in Social Networks: A Multi-Stage Approach

Contact Info

Product

Resources

About