Supervised learning for suicidal ideation detection in online user content

Shaoxiong Ji, Celina Ping Yu, Sai-Fu Fung, Shirui Pan, Guodong Long

Research output: Contribution to journalArticleResearchpeer-review

3 Citations (Scopus)

Abstract

Early detection and treatment are regarded as the most effective ways to prevent suicidal ideation and potential suicide attempts-two critical risk factors resulting in successful suicides. Online communication channels are becoming a new way for people to express their suicidal tendencies. This paper presents an approach to understand suicidal ideation through online user-generated content with the goal of early detection via supervised learning. Analysing users' language preferences and topic descriptions reveals rich knowledge that can be used as an early warning system for detecting suicidal tendencies. Suicidal individuals express strong negative feelings, anxiety, and hopelessness. Suicidal thoughts may involve family and friends. And topics they discuss cover both personal and social issues. To detect suicidal ideation, we extract several informative sets of features, including statistical, syntactic, linguistic, word embedding, and topic features, and we compare six classifiers, including four traditional supervised classifiers and two neural network models. An experimental study demonstrates the feasibility and practicability of the approach and provides benchmarks for the suicidal ideation detection on the active online platforms: Reddit SuicideWatch and Twitter.

Original languageEnglish
Article number6157249
Number of pages10
JournalComplexity
Volume2018
DOIs
Publication statusPublished - 2018
Externally publishedYes

Cite this

Ji, Shaoxiong ; Yu, Celina Ping ; Fung, Sai-Fu ; Pan, Shirui ; Long, Guodong. / Supervised learning for suicidal ideation detection in online user content. In: Complexity. 2018 ; Vol. 2018.
@article{cc7bcc6c71fc491abda91e28ad6c76dd,
title = "Supervised learning for suicidal ideation detection in online user content",
abstract = "Early detection and treatment are regarded as the most effective ways to prevent suicidal ideation and potential suicide attempts-two critical risk factors resulting in successful suicides. Online communication channels are becoming a new way for people to express their suicidal tendencies. This paper presents an approach to understand suicidal ideation through online user-generated content with the goal of early detection via supervised learning. Analysing users' language preferences and topic descriptions reveals rich knowledge that can be used as an early warning system for detecting suicidal tendencies. Suicidal individuals express strong negative feelings, anxiety, and hopelessness. Suicidal thoughts may involve family and friends. And topics they discuss cover both personal and social issues. To detect suicidal ideation, we extract several informative sets of features, including statistical, syntactic, linguistic, word embedding, and topic features, and we compare six classifiers, including four traditional supervised classifiers and two neural network models. An experimental study demonstrates the feasibility and practicability of the approach and provides benchmarks for the suicidal ideation detection on the active online platforms: Reddit SuicideWatch and Twitter.",
author = "Shaoxiong Ji and Yu, {Celina Ping} and Sai-Fu Fung and Shirui Pan and Guodong Long",
year = "2018",
doi = "10.1155/2018/6157249",
language = "English",
volume = "2018",
journal = "Complexity",
issn = "1076-2787",
publisher = "John Wiley & Sons",

}

Supervised learning for suicidal ideation detection in online user content. / Ji, Shaoxiong; Yu, Celina Ping; Fung, Sai-Fu; Pan, Shirui; Long, Guodong.

In: Complexity, Vol. 2018, 6157249, 2018.

Research output: Contribution to journalArticleResearchpeer-review

TY - JOUR

T1 - Supervised learning for suicidal ideation detection in online user content

AU - Ji, Shaoxiong

AU - Yu, Celina Ping

AU - Fung, Sai-Fu

AU - Pan, Shirui

AU - Long, Guodong

PY - 2018

Y1 - 2018

N2 - Early detection and treatment are regarded as the most effective ways to prevent suicidal ideation and potential suicide attempts-two critical risk factors resulting in successful suicides. Online communication channels are becoming a new way for people to express their suicidal tendencies. This paper presents an approach to understand suicidal ideation through online user-generated content with the goal of early detection via supervised learning. Analysing users' language preferences and topic descriptions reveals rich knowledge that can be used as an early warning system for detecting suicidal tendencies. Suicidal individuals express strong negative feelings, anxiety, and hopelessness. Suicidal thoughts may involve family and friends. And topics they discuss cover both personal and social issues. To detect suicidal ideation, we extract several informative sets of features, including statistical, syntactic, linguistic, word embedding, and topic features, and we compare six classifiers, including four traditional supervised classifiers and two neural network models. An experimental study demonstrates the feasibility and practicability of the approach and provides benchmarks for the suicidal ideation detection on the active online platforms: Reddit SuicideWatch and Twitter.

AB - Early detection and treatment are regarded as the most effective ways to prevent suicidal ideation and potential suicide attempts-two critical risk factors resulting in successful suicides. Online communication channels are becoming a new way for people to express their suicidal tendencies. This paper presents an approach to understand suicidal ideation through online user-generated content with the goal of early detection via supervised learning. Analysing users' language preferences and topic descriptions reveals rich knowledge that can be used as an early warning system for detecting suicidal tendencies. Suicidal individuals express strong negative feelings, anxiety, and hopelessness. Suicidal thoughts may involve family and friends. And topics they discuss cover both personal and social issues. To detect suicidal ideation, we extract several informative sets of features, including statistical, syntactic, linguistic, word embedding, and topic features, and we compare six classifiers, including four traditional supervised classifiers and two neural network models. An experimental study demonstrates the feasibility and practicability of the approach and provides benchmarks for the suicidal ideation detection on the active online platforms: Reddit SuicideWatch and Twitter.

UR - http://www.scopus.com/inward/record.url?scp=85062845499&partnerID=8YFLogxK

U2 - 10.1155/2018/6157249

DO - 10.1155/2018/6157249

M3 - Article

VL - 2018

JO - Complexity

JF - Complexity

SN - 1076-2787

M1 - 6157249

ER -