La qualité de l'information dans les réseaux sociaux en ligne : une approche non supervisée et rapide de détection de spam

Online social networks provide data valuable for a tremendous range of applications. However, the easy-to-use interfaces and low limits of the publication generate various information quality problems, such the user-generated content in such networks. The existence of ill-intentioned users, so-called social spammers, belongs to this environments. The major limitations of the methods detecting spams are the use of supervised learning approaches that requiring ground truth data-sets. Moreover, the account-based detection methods are not practical for processing "crawled" large collections of social posts, requiring months to process such collections. Hence, in this paper, we introduce a design of an unsupervised learning approach dedicated for detecting spam accounts existing in large collections of "trending" topics. Our experimental evaluation on Twitter demonstrates the efficiency of our approach as well as its speed comparing to supervised approaches. MOTS-CLÉS : Twitter, Réseau Social, Spam