Transferable Natural Language Interface to Structured Queries aided by Adversarial Generation

TL;DR

This paper introduces a transfer learning approach for natural language interfaces to structured queries, combining a schema-aware neural model and adversarial data augmentation to improve domain adaptation with limited target data.

Contribution

It proposes SQIN, a schema-aware neural model for better domain transfer, and AugmentGAN, an adversarial data augmentation method, achieving state-of-the-art results.

Findings

State-of-the-art performance on GeoQuery, Overnight, and WikiSQL datasets.

Effective domain transfer with limited target domain data.

Enhanced SQL decoding with schema-awareness.

Abstract

A natural language interface (NLI) to structured query is intriguing due to its wide industrial applications and high economical values. In this work, we tackle the problem of domain adaptation for NLI with limited data on target domain. Two important approaches are considered: (a) effective general-knowledge-learning on source domain semantic parsing, and (b) data augmentation on target domain. We present a Structured Query Inference Network (SQIN) to enhance learning for domain adaptation, by separating schema information from NL and decoding SQL in a more structural-aware manner; we also propose a GAN-based augmentation technique (AugmentGAN) to mitigate the issue of lacking target domain data. We report solid results on GeoQuery, Overnight, and WikiSQL to demonstrate state-of-the-art performances for both in-domain and domain-transfer tasks.