StructVAE: Tree-structured Latent Variable Models for Semi-supervised Semantic Parsing

  • 2018-06-20 16:29:01
  • Pengcheng Yin, Chunting Zhou, Junxian He, Graham Neubig
  • 54

Abstract

Semantic parsing is the task of transducing natural language (NL) utterancesinto formal meaning representations (MRs), commonly represented as treestructures. Annotating NL utterances with their corresponding MRs is expensiveand time-consuming, and thus the limited availability of labeled data oftenbecomes the bottleneck of data-driven, supervised models. We introduceStructVAE, a variational auto-encoding model for semisupervised semanticparsing, which learns both from limited amounts of parallel data, andreadily-available unlabeled NL utterances. StructVAE models latent MRs notobserved in the unlabeled data as tree-structured latent variables. Experimentson semantic parsing on the ATIS domain and Python code generation show thatwith extra unlabeled data, StructVAE outperforms strong supervised models.

 

Quick Read (beta)

loading the full paper ...