Learning to perform moderation in online forums

Online discussion forums are a valuable resource for people looking to find information, discuss ideas, and get advice on the Internet. Unfortunately, many forums have too much activity and information available, resulting in information overload. Moderation systems are implemented in some forums as a way to handle this problem, but due to sparsity issues, they are often not sufficient. We describe a novel method for learning from past moderations to develop a classifier that can perform automated moderation and thus address the sparsity problem. Additionally, we discuss the possibility of training a moderating classifier on a moderated forum and then applying it to an otherwise unmoderated forum.