Hidden object-ranking problems (HORPs) are object-ranking problems stated in instance-ranking terms. To our knowledge, there is no algorithm able to process these problems with the appropriate bias. This lack is not significant as long as the size of the dataset makes it possible to capture enough information by mining more data; however, when the data are scarce, any information lying in the data is worth exploiting and such an algorithm would become useful. We explicit the appropriate bias for object-ranking problems and propose an algorithm able to apply this bias to cases where these problems arise in an instance-ranking form. The theoretical foundations of the algorithm are discussed and the algorithm is tested on scarce real data, yielding better results (94.4% accuracy) than traditional algorithms (92.6% accuracy for the best case).