Coarse-to-Fine Decoding for Neural Semantic Parsing
This addresses the problem of mapping natural language to structured meaning representations for applications like question answering or database queries, but it is incremental as it builds on existing neural parsing methods.
The paper tackles semantic parsing by proposing a two-stage coarse-to-fine neural architecture that first generates a rough sketch of meaning and then fills in details, achieving competitive results across four datasets with different domains and representations.
Semantic parsing aims at mapping natural language utterances into structured meaning representations. In this work, we propose a structure-aware neural architecture which decomposes the semantic parsing process into two stages. Given an input utterance, we first generate a rough sketch of its meaning, where low-level information (such as variable names and arguments) is glossed over. Then, we fill in missing details by taking into account the natural language input and the sketch itself. Experimental results on four datasets characteristic of different domains and meaning representations show that our approach consistently improves performance, achieving competitive results despite the use of relatively simple decoders.