Large-Scale Vandalism Detection with Linear Classifiers - The Conkerberry Vandalism Detector at WSDM Cup 2017

• DOI: 10.48550/arxiv.1712.06920
• Publication Date: 2017-01-01
• Authors: А. В. Григорьев
• Keywords: Offensive

Nowadays many artificial intelligence systems rely on knowledge bases for enriching the information they process. Such Knowledge Bases are usually difficult to obtain and therefore crowdsourced: available everyone internet suggest edits add new information. Unfortunately, sometimes targeted by vandals who put inaccurate or offensive there. This is especially bad that use these Bases: them it important reliable make correct inferences. One of such Wikidata, fight organizers WSDM Cup 2017 challenged participants build a model detecting mistrustful edits. In this paper we present second place solution cup: show possible achieve competitive performance with simple linear classification. With our approach can AU ROC 0.938 test data. Additionally, compared other approaches, ours significantly faster. The made GitHub.