This study addresses the challenge of triaging posts in mental health forums, where users often express severe distress and risk of self-harm. To help moderators respond promptly to critical posts, the authors propose a classification approach that sorts content into four severity categories based on self-harm ideation. Using a feature-rich framework incorporating lexical, psycholinguistic, contextual, and topic modeling features, the proposed models significantly improve content severity triaging (up to 17% better F-1 scores than previous methods). The study also shows that long-term forum users tend to exhibit decreased risk severity over time, and highlights the difficulty moderators face in identifying critical posts without automated tools.
-
Cohan, Arman, et al. "Triaging content severity in online mental health forums." Journal of the Association for Information Science and Technology 68.11 (2017): 2675-2689.
-
Members
- abobader
- Kirill Gromov
- V0RT3X
- IC Essentials
- JoelR
- envy
- Analog
- Sinistra
- DawPi
- eivindsimensen
- Dilip
- Square Wheels
- N700
- Majster87
- Adriano Faria
- opentype
- adik
- bernhara
- Como
- GrantHorizons
- A Zayed
- terabyte
- Videoflicks
- Voyage
- burnyourfeelings
- onlyME
- Jon Erickson
- Chris Anderson
- aXenDev
- AnonDoggo
- Empire
- Nathan Explosion
- Labis
- XwReK
- COSMIN
- Hexzon
- IPS THEME
- Venthas
- TomCat
- ReyDev
- StevenM
- GazzaGarratt
- Live Games
- Charlie Feigel
- Auto Evoke
- Steph40
- JoeyM
- Ryan
- master963
- John Horton
- TwinTurbo
- Uncrowned Gaurd
- Foxtrek64
- Claudia999
- bing11
- Andy Y
- Copycat
- Karyexo Karyexooo
- Kelkrel
- Myr
- dolphin
- ali hagi
- Richard Arch
- yaotzin
- lanc3lot
- Brian
- Nicolas PC
- Synergy
- Maria
- Nomad
- TheLlamaman
- scaz
- The SoftBay
- TracyIsland
- Yurii
- william trowbridge
- Cristian Croitor
- Asare
- djdan36
- shahed
- YalcinA
- VAHID
- send2yoni
- ZLTRGO
- Paul Kaiser
- Paul
- Omar Barbeytia carretero
- Ryancoolround
- rainx
- YourSharona
- Kentraiyle Robinson
- MichaelR
- Edward Ellas
- PrettyPixels
- Denis Dyack
- DursunKaptan
- MissB
- aLEX49566
- Codepixel
- alsl sndnxnx