Improving Quality and Efficiency in Plan-based Neural Data-to-Text Generation

TL;DR

This paper enhances neural data-to-text generation by introducing a trainable planner, typing hints, verification reranking, and referring expression modules, resulting in faster, more fluent, and more faithful text outputs.

Contribution

It presents four novel extensions to the existing step-by-step neural data-to-text framework, significantly improving efficiency and output quality.

Findings

Generation speed increased by orders of magnitude

Improved faithfulness through reranking

Enhanced handling of unseen relations and entities

Abstract

We follow the step-by-step approach to neural data-to-text generation we proposed in Moryossef et al (2019), in which the generation process is divided into a text-planning stage followed by a plan-realization stage. We suggest four extensions to that framework: (1) we introduce a trainable neural planning component that can generate effective plans several orders of magnitude faster than the original planner; (2) we incorporate typing hints that improve the model's ability to deal with unseen relations and entities; (3) we introduce a verification-by-reranking stage that substantially improves the faithfulness of the resulting texts; (4) we incorporate a simple but effective referring expression generation module. These extensions result in a generation process that is faster, more fluent, and more accurate.