Predicting Controversy of Wikipedia Articles Using the Article Feedback Tool

Different points of view, opinions and controversies constitute the inherent part of modern society. Early detection of controversy is crucial for increasing productivity in peer production systems. The paper presents novelty approach to detecting controversial articles on the Wikipedia based on users ratings from Article Feedback Tool. The performance of proposed approach is on par with state-of-the-art solutions, but may also be applied outside Wikipedia-like systems. Additionally, emotion polarity measures can be used to locate controversial parts of articles, based on talk pages sections. With help of proposed algorithms, all articles in English Wikipedia have been tagged as either controversial or non-controversial. The dataset has been published and can be used by other researchers.