Spam is an enemy of our resources and mood. In this paper spam detection is examined in case of social media content. In particular a typical Bayesian classifier is setup and trained on real data. Additionally a social computing method is also introduced, which analyses the attention that content receives. Finally a novel hybrid approach is implemented, where content is characterized as spam when both the Bayesian classifier and the social attention model agree on content’s evaluation. Experimental results on real world Twitter content, exhibit the promising performance of the proposed scheme.