Dataflow Dialogue Generation

TL;DR

This paper explores dataflow-based task-oriented dialogue generation, demonstrating improved translation accuracy through dialogue augmentation in two domains, MultiWOZ and SMCalFlow.

Contribution

It introduces a dataflow dialogue paradigm and shows how dialogue generation can enhance data translation accuracy in task-oriented systems.

Findings

Improved translation accuracy with dialogue augmentation

Effective dataflow dialogue generation in multiple domains

Demonstrated benefits of agenda-driven and non-agenda approaches

Abstract

We demonstrate task-oriented dialogue generation within the dataflow dialogue paradigm. We show an example of agenda driven dialogue generation for the MultiWOZ domain, and an example of generation without an agenda for the SMCalFlow domain, where we show an improvement in the accuracy of the translation of user requests to dataflow expressions when the generated dialogues are used to augment the translation training dataset.