In this paper, we propose a new method for characterizing bytecode-only smart contracts by automatically assigning multiple attribute tags. Using a deep learning approach, our system, the ScanAT, ...