Exploiting document structure and feature hierarchy for semi-supervised domain adaptation

In this work we try to bridge the gap often encountered by researchers who find themselves with few or no labeled examples from their desired target domain, yet still have access to large amounts of l