Mining Twitter for Suicide Prevention

Abboute, Amayas; Boudjeriou, Yasser; Entringer, Gilles; Azé, Jérôme; Bringay, Sandra; Poncelet, Pascal

doi:10.1007/978-3-319-07983-7_36

Cited by 57 publications

(38 citation statements)

References 1 publication

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…This dataset, which may be considered as a longitudinal data, was collected from 10 unique real users that demonstrated a serious change in speech or online behavior in their Twitter accounts. Two initial validation cases, which ended with the individual committing suicide, were previously identified by [40]. We were able to identify a third fatal case and an additional seven other cases where the individuals demonstrated an abrupt change in behavior.…”

Section: Experimental Evaluation Data Collection and Annotationmentioning

confidence: 93%

Detection of suicide-related posts in Twitter data streams

Vioulès¹,

Moulahi²,

Azé³

et al. 2018

IBM J. Res. & Dev.

120

View full text Add to dashboard Cite

Suicidal ideation detection in online social networks is an emerging research area with major challenges. Recent research has shown that the publicly available information spread across social media platforms holds valuable indicators to effectively detecting individuals with suicidal intentions. The key challenge of suicide prevention is understanding and detecting the complex risk factors and warning signs that may precipitate the event. In this paper, we present a new approach that uses the social media platform Twitter to quantify suicide-warning signs for individuals and to detect posts containing suicide-related content. The main originality of this approach is the automatic identification of sudden changes in a user's online behavior. To detect such changes, we combine natural language processing techniques to aggregate behavioral and textual features and pass these features through a martingale framework, which is widely used for change detection in data streams. Experiments show that our text-scoring approach effectively captures warning signs in text compared to traditional machine learning classifiers. Additionally, the application of the martingale framework highlights changes in online behavior and shows promise for detecting behavioral changes in at-risk individuals.

show abstract

Section: Experimental Evaluation Data Collection and Annotationmentioning

confidence: 93%

Detection of suicide-related posts in Twitter data streams

Vioulès¹,

Moulahi²,

Azé³

et al. 2018

IBM J. Res. & Dev.

120

View full text Add to dashboard Cite

show abstract

“…Nosúltimos anos, diversos trabalhos investigaram a possibilidade de criar modelos para a classificação de conteúdo relacionado ao suicídio nas redes sociais. Em [Abboute et al 2014], foi descrito um processo completo para, automaticamente, coletar tweets suspeitos de acordo com um vocabulário de termos, criado pelos próprios autores, que pessoas suicidas costumam utilizar. Com a obtenção de um corpus que também incluía casos comprovados, os tweets foram classificados em "risky" e "non risky".…”

Section: Trabalhos Relacionadosunclassified

Detecting Suicidal Ideation on Tweets

Cardoso¹,

Sinoara²,

Rezende³

et al. 2019

Anais Do Encontro Nacional De Inteligência Artificial E Computacional (ENIAC 2019)

View full text Add to dashboard Cite

According to the World Health Organization, every 40s a person dies of suicide in the world. Among young people aged 15 to 29, suicide is the second largest cause of death. Yet such deaths can be prevented. In this scenario, social networks like Twitter can become sources of information in real time and help in suicide prevention. The present work makes an initial exploration of the problem of identifying individuals at risk of suicide in social networks in Portuguese Language. As a main result, a manually labeled dataset of tweets has been constructed that can be used in future research on the subject.

show abstract

“…The variation here could suggest the flippant use of such phrases on social media when having a bad day -hence the additional challenges posed to classification of suicidal ideation on social media. Finally, [1] used machine learning to classify 'risky' and 'non risky' Tweets, as defined by human annotators, with an accuracy of around 60%. They created word lists to represent a number of topics and emotions related to suicide, finding references to insults, hurt and bullying in the 'risky' category.…”

Section: Related Workmentioning

confidence: 99%

“…These were incorporated because of the particularly emotive nature of the task. Emotions such as fear, anger and general aggressiveness are particularly prominent in suicidal communication [1] • Features representing ideosyncratic language expressed in short, informal text such as social media posts within a limited number of characters. These were extracted from the annotated Tumblr posts we collected to try and incorporate the language used on social media that may not be identified using standard text mining features.…”

Section: Feature Preparationmentioning

confidence: 99%

“…Suicidal language is likely to include emotive content and possible signs of depression but we do not suggest depression and suicidal ideation are synonymous in this paper. Two very recent papers presented the results of Twitter studies aiming to classify 'risky' language [1] and levels of 'distress' [16] both reporting classification performance that has potential for improvement (around 60-64%). An important step in providing support to suicidal social media users is to understand how suicidal ideation is communicated.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Machine Classification and Analysis of Suicide-Related Communication on Twitter

Burnap

Colombo

Scourfield

2015

Proceedings of the 26th ACM Conference on Hypertext &Amp; Social Media - HT '15

128

140

View full text Add to dashboard Cite

The World Wide Web, and online social networks in particular, have increased connectivity between people such that information can spread to millions of people in a matter of minutes. This form of online collective contagion has provided many benefits to society, such as providing reassurance and emergency management in the immediate aftermath of natural disasters. However, it also poses a potential risk to vulnerable Web users who receive this information and could subsequently come to harm. One example of this would be the spread of suicidal ideation in online social networks, about which concerns have been raised. In this paper we report the results of a number of machine classifiers built with the aim of classifying text relating to suicide on Twitter. The classifier distinguishes between the more worrying content, such as suicidal ideation, and other suicide-related topics such as reporting of a suicide, memorial, campaigning and support. It also aims to identify flippant references to suicide. We built a set of baseline classifiers using lexical, structural, emotive and psychological features extracted from Twitter posts. We then improved on the baseline classifiers by building an ensemble classifier using the Rotation Forest algorithm and a Maximum Probability voting classification decision method, based on the outcome of base classifiers. This achieved an F-measure of 0.728 overall (for 7 classes, including suicidal ideation) and 0.69 for the suicidal ideation class. We summarise the results by reflecting on the most significant predictive principle components of the suicidal ideation class to provide insight into the language used on Twitter to express suicidal ideation.

show abstract

Mining Twitter for Suicide Prevention

Cited by 57 publications

References 1 publication

Detection of suicide-related posts in Twitter data streams

Detection of suicide-related posts in Twitter data streams

Detecting Suicidal Ideation on Tweets

Machine Classification and Analysis of Suicide-Related Communication on Twitter

Contact Info

Product

Resources

About