Parsing NLmaps Queries Using Adversarial Neural Machine Translation

In contrast to “traditional” Neural Machine Translation, Adversarial
NMT (Wu, Lijun et al. 2017; Yang, Zhen et al. 2018) uses a Generative
Adversarial Networks approach to train a translation network together
with a network that estimates a translation’s quality. In my
Bachelor’s thesis, I adapted this approach to neural semantic parsing
and applied it to the NLmaps dataset (Haas and Riezler 2016). The
performance of ANMT is compared with the latest results on the NLmaps
dataset, which were produced by Lawrence and Riezler (2018).