A Split-and-Recombine Approach for Follow-up Query Analysis

TL;DR

This paper introduces STAR, a parser-independent two-phase approach for analyzing follow-up queries in context-dependent semantic parsing, significantly improving accuracy across multiple datasets and scenarios.

Contribution

The paper presents STAR, a novel split-and-recombine method that effectively handles diverse follow-up query scenarios without relying on specific parsers.

Findings

STAR outperforms baseline by nearly 8% on FollowUp dataset.

The approach is effective across different domains and datasets.

Extensible to other datasets like SQA.

Abstract

Context-dependent semantic parsing has proven to be an important yet challenging task. To leverage the advances in context-independent semantic parsing, we propose to perform follow-up query analysis, aiming to restate context-dependent natural language queries with contextual information. To accomplish the task, we propose STAR, a novel approach with a well-designed two-phase process. It is parser-independent and able to handle multifarious follow-up scenarios in different domains. Experiments on the FollowUp dataset show that STAR outperforms the state-of-the-art baseline by a large margin of nearly 8%. The superiority on parsing results verifies the feasibility of follow-up query analysis. We also explore the extensibility of STAR on the SQA dataset, which is very promising.