Improving Semantic Parsing for Task Oriented Dialog

TL;DR

This paper enhances semantic parsing for task-oriented dialogue by integrating contextualized embeddings, ensembling, and re-ranking, significantly improving accuracy and reducing errors on the TOP dataset.

Contribution

It introduces three novel improvements to hierarchical semantic parsing models, achieving state-of-the-art results on the TOP dataset.

Findings

6.4% increase in exact match accuracy

33% reduction in parsing errors

Effective correction of different error types

Abstract

Semantic parsing using hierarchical representations has recently been proposed for task oriented dialog with promising results [Gupta et al 2018]. In this paper, we present three different improvements to the model: contextualized embeddings, ensembling, and pairwise re-ranking based on a language model. We taxonomize the errors possible for the hierarchical representation, such as wrong top intent, missing spans or split spans, and show that the three approaches correct different kinds of errors. The best model combines the three techniques and gives 6.4% better exact match accuracy than the state-of-the-art, with an error reduction of 33%, resulting in a new state-of-the-art result on the Task Oriented Parsing (TOP) dataset.