We evaluate the character-level translation method for neural semantic parsing on a large corpus of sentences annotated with Abstract Meaning Representations (AMRs). Using a seq2seq model, and some trivial preprocessing and postprocessing of AMRs, we obtain a baseline accuracy of 53.1 (F-score on AMR-triples). We examine four different approaches to improve this baseline result: (i) reordering AMR branches to match the word order of the input sentence increases performance to 58.3; (ii) adding part-of-speech tags (automatically produced) to the input shows improvement as well (57.2); (iii) So does the introduction of super characters (conflating frequent sequences of characters to a single character), reaching 57.4; (iv) adding silver-standard training data obtained by an off-the-shelf parser yields the biggest improvement, resulting in an F-score of 64.0. Combining all four techniques leads to an F-score of 69.0, which is state-of-the-art in AMR parsing. This is remarkable because of the relatively simplicity of the approach: the only explicit linguistic knowledge that we use are part-of-speech tags.

我们使用字符级翻译的方法对一大语料库中注释有抽象意义表示的句子进行神经语义解析的评估。通过一些简单的预处理和后处理，我们使用序列到序列模型得到了一个基线准确率为53.1。研究了五种不同的方法来改善这个基线结果，结合这五种技术的结果使获得了一项AMR解析的最新进展状态的成果，F分数为71.0。

基于字符级别翻译的神经语义解析: 对抽象意义表示的实验