This paper discusses collaborative filtering systems in online communities, which aim to manage information overload and improve discourse by allowing users to rank content based on validity or usefulness. However, these systems have known weaknesses, such as dependency on a large crowd for ratings and concerns about representativeness. The study proposes using statistical machine learning to predict community ratings, extracting features from metadata and linguistic content to identify predictors of community ratings. Using Slashdot as an example, the study finds that author reputation, use of pronouns, and author sentiment are significant predictors, achieving 76% accuracy in predicting community ratings.
-
Brennan, Michael Robert, Stacey Wrazien, and Rachel Greenstadt. "Learning to Extract Quality Discourse in Online Communities." Collaboratively-Built Knowledge Sources and AI. 2010.
-
Members
- Voyage
- GazzaGarratt
- rainx
- Adriano Faria
- opentype
- Dilip
- Nathan Explosion
- YourSharona
- JoelR
- DawPi
- Live Games
- bernhara
- Kentraiyle Robinson
- IC Essentials
- Myr
- Chris Anderson
- Como
- Brian
- MichaelR
- Edward Ellas
- Square Wheels
- IPS THEME
- aXenDev
- PrettyPixels
- Analog
- Denis Dyack
- A Zayed
- Labis
- DursunKaptan
- V0RT3X
- MissB
- TheLlamaman
- Andy Y
- envy
- eivindsimensen
- JoeyM
- aLEX49566
- adik
- Codepixel
- alsl sndnxnx
- burnyourfeelings
- N700
- isvans
- Marius
- Matt
- terabyte
- Thomas Taschler
- Surpac
- JoshB
- Ioannis D
- abobader
- ReyDev
- Richard Arch
- bdmusic 24
- Majster87
- TomCat
- Paul
- Pmw
- Torgeir Rui
- Kammer et
- Nicolas PC
- XwReK
- Claudia999
- Kirill Gromov
- Synergy
- bing11
- Marvin Moore
- ArashDev
- onlyME
- ali hagi
- StevenM
- NewVicious
- lukash
- TracyIsland
- Andhrafriends Admin
- Daffy
- hyprem
- GuitarGathering
- Tripp
- Askancy
- MLK
- Jelly Belly
- eveneme eveneme
- Nomad
- Morphe
- lordi
- shahed
- John Horton
- PayMap
- Serval
- Nomer3
- Dennis Maidon
- Zennuie
- COSMIN
- wulfx01
- Matthew Hawley
- Verto
- George Anderssen
- Toby
- Cheryl