Despite the essential role of microRNAs in post-transcriptional regulation, their genes and related mechanisms are still elusive. There is a wide variety of experimental and in silico methods for target exploration, prediction and validation, but these methods are somewhat complementary and suffer from different biases. Because the profile of microRNA targets allows the definition of similarity over miRNAs, these different information sources result in highly heterogeneous similarities. We investigate the adoption of multiple kernel learning framework for the fusion of these similarities, specifically we apply it for miRNA prioritization in the one-class settings. We use the described methodology in breast cancer, illustrate its technological aspects and validate our results not just by the common leave-one-out cross validation, but with prospective evaluation as well.