Machine Learning Based Detection of Vandalism in Wikipedia across Languages

Applying machine learning algorithms for detecting vandalism in two languages are described in this paper. Vandalism is a major issue in Wikipedia as it accounts for about 1% of edits during 2015. The majority of vandalism is from human editors, whose vandalism can be traced through access and edit logs. In this paper, we propose using a list of classifiers in one language, and then evaluate them across languages in two datasets: the hourly count of views of each Wikipedia article, and the used edit history of articles. For this purpose, Simple English and Albanian Wikipedia datasets will be used. The results obtained show that the characteristic features of vandalism can be learned from view and edit patterns, and models built in one language can be applied successfully to other languages.