A computer-implemented method for classifying an input data set within a data category using multiple data representation modes. The method includes identifying at least a first data representation source mode and a second data representation source mode; classifying the at least first data representation source mode via at least a first data recognition tool and the at least second data representation source mode via at least a second data recognition tool, the classifying including allocating a confidence factor for each data representation source mode in the data category; and combining outputs of the classifying into a single output confidence score by using a weighted fusion of the allocated confidence factors.