Automatic Moderation of Comments in a Large On-line Journalistic Environment

On-line journalistic sites publish several news and stories every day. Readers of these sites may comment a story, and, as a consequence, a single story might receive thousands of comments. The quality of these comments may vary a lot, from spams to truly useful information. Separating good from bad comments is the primary goal of comment moderation. In this paper we address the problem of automatic moderation of comments in a large journalistic Web site. Participants of the site may interact with each other, constituting a large social network. We propose a classification technique which combines underlying implicit patterns in the comments’ content with patterns hidden in the social network, and then uses the result for automatic moderation. We evaluate our proposed technique using a real collection of comments collected from the Slashdot forum. We compared the proposed technique against traditional ones, such as decision trees and SVMs. We observed that the proposed technique is very effective for scoring comments, reaching more than 96% of accuracy. Classifying comments seems to be a more complex task, and the proposed technique achieves almost 67% of accuracy. Further, the proposed technique is very fast, being able to classify and score hundreds of comments per minute.