Conversational Question Reformulation via Sequence-to-Sequence Architectures and Pretrained Language Models

TL;DR

This paper empirically evaluates pretrained language models, especially T5, for conversational question reformulation, demonstrating their effectiveness in improving task-oriented dialogue systems on multiple benchmarks.

Contribution

It introduces an empirical study of PLMs for CQR, highlighting T5's superior performance with fewer parameters across in-domain and out-domain datasets.

Findings

T5 achieves the best results on CANARD and CAsT datasets.

Pretrained models outperform traditional sequence-to-sequence architectures.

Fewer parameters are needed for optimal performance with T5.

Abstract

This paper presents an empirical study of conversational question reformulation (CQR) with sequence-to-sequence architectures and pretrained language models (PLMs). We leverage PLMs to address the strong token-to-token independence assumption made in the common objective, maximum likelihood estimation, for the CQR task. In CQR benchmarks of task-oriented dialogue systems, we evaluate fine-tuned PLMs on the recently-introduced CANARD dataset as an in-domain task and validate the models using data from the TREC 2019 CAsT Track as an out-domain task. Examining a variety of architectures with different numbers of parameters, we demonstrate that the recent text-to-text transfer transformer (T5) achieves the best results both on CANARD and CAsT with fewer parameters, compared to similar transformer architectures.