UCL+Sheffield at SemEval-2016 Task 8: Imitation learning for AMR parsing with an alpha-bound

22405

Knowledge Triples