Models and Benchmarks for Representation Learning of Partially Observed Subgraphs