Improving Quality and Efficiency in Plan-based Neural Data-to-Text Generation
This work addresses data-to-text generation for natural language processing applications, presenting incremental improvements to an existing framework.
The authors tackled the problem of neural data-to-text generation by extending a step-by-step framework to improve speed, fluency, and accuracy, resulting in a trainable planner that is orders of magnitude faster and a verification stage that substantially enhances faithfulness.
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.