Saved in:
| Main Authors: | , |
|---|---|
| Format: | Preprint |
| Published: |
2023
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2310.06056 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866914673198104576 |
|---|---|
| author | Dus, Yasin Nefedov, Georgiy |
| author_facet | Dus, Yasin Nefedov, Georgiy |
| contents | The World Health Organization (WHO) estimated that approximately 1.4 million individuals worldwide died by suicide in 2022. This figure indicates that one person died by suicide every 20 s during the year. Globally, suicide is the tenth-leading cause of death, while it is the second-leading cause of death among young people aged 15329 years. In 2022, it was estimated that approximately 10.5 million suicide attempts would occur. The WHO suggests that along with each completed suicide attempt, many individuals attempt suicide. Today, social media is a place in which people share their feelings. Thus, social media can help us understand the thoughts and possible actions of individuals. This study leverages this advantage and focuses on developing an automated model to use information from social media to determine whether someone is contemplating self-harm. This model is based on the Suicidal-ELECTRA model. We collected datasets of social media posts, processed them, and used them to train and fiune-tune our model. Evaluation of the refined model with a testing dataset consistently yielded outstanding results. The model had an impressive accuracy rate of 93% and commendable F1 score of 0.93. Additionally, we developed an application programming interface that seamlessly integrated our tool with third-party platforms, enhancing its implementation potential to address the concern of rising suicide rates. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2310_06056 |
| institution | arXiv |
| publishDate | 2023 |
| record_format | arxiv |
| spellingShingle | An Automated Tool to Detect Suicidal Susceptibility from Social Media Posts Dus, Yasin Nefedov, Georgiy Computers and Society K.4.2 The World Health Organization (WHO) estimated that approximately 1.4 million individuals worldwide died by suicide in 2022. This figure indicates that one person died by suicide every 20 s during the year. Globally, suicide is the tenth-leading cause of death, while it is the second-leading cause of death among young people aged 15329 years. In 2022, it was estimated that approximately 10.5 million suicide attempts would occur. The WHO suggests that along with each completed suicide attempt, many individuals attempt suicide. Today, social media is a place in which people share their feelings. Thus, social media can help us understand the thoughts and possible actions of individuals. This study leverages this advantage and focuses on developing an automated model to use information from social media to determine whether someone is contemplating self-harm. This model is based on the Suicidal-ELECTRA model. We collected datasets of social media posts, processed them, and used them to train and fiune-tune our model. Evaluation of the refined model with a testing dataset consistently yielded outstanding results. The model had an impressive accuracy rate of 93% and commendable F1 score of 0.93. Additionally, we developed an application programming interface that seamlessly integrated our tool with third-party platforms, enhancing its implementation potential to address the concern of rising suicide rates. |
| title | An Automated Tool to Detect Suicidal Susceptibility from Social Media Posts |
| topic | Computers and Society K.4.2 |
| url | https://arxiv.org/abs/2310.06056 |