Document-Level Text Generation with Minimum Bayes Risk Decoding using Optimal Transport

TL;DR

This paper introduces MBR-OT, a novel adaptation of Minimum Bayes Risk decoding using Wasserstein distance, to improve document-level text generation by better capturing the utility of longer contexts, outperforming standard MBR.

Contribution

We propose MBR-OT, a new method that applies optimal transport to enhance utility estimation in document-level text generation tasks.

Findings

MBR-OT outperforms standard MBR in multiple tasks

Wasserstein distance effectively captures document utility

Improved performance in translation, simplification, and captioning

Abstract

Document-level text generation tasks are known to be more difficult than sentence-level text generation tasks as they require the understanding of longer context to generate high-quality texts. In this paper, we investigate the adaption of Minimum Bayes Risk (MBR) decoding for document-level text generation tasks. MBR decoding makes use of a utility function to estimate the output with the highest expected utility from a set of candidate outputs. Although MBR decoding is shown to be effective in a wide range of sentence-level text generation tasks, its performance on document-level text generation tasks is limited as many of the utility functions are designed for evaluating the utility of sentences. To this end, we propose MBR-OT, a variant of MBR decoding using Wasserstein distance to compute the utility of a document using a sentence-level utility function. The experimental result shows that the performance of MBR-OT outperforms that of the standard MBR in document-level machine translation, text simplification, and dense image captioning tasks. Our code is available at https://github.com/jinnaiyuu/mbr-optimal-transport