MSDF: A General Open-Domain Multi-Skill Dialog Framework

TL;DR

This paper introduces MSDF, a versatile multi-skill dialog framework that leverages pre-trained models and external knowledge to excel across various dialog tasks, demonstrating superior performance and competitiveness.

Contribution

The paper presents a novel general dialog framework with a transferable response generator, consistency selector, and external knowledge integration, applicable across multiple dialog scenarios.

Findings

MSDF outperforms baseline models significantly.

MSDF won 3rd prize in a major dialog challenge.

The framework effectively handles diverse dialog tasks.

Abstract

Dialog systems have achieved significant progress and have been widely used in various scenarios. The previous researches mainly focused on designing dialog generation models in a single scenario, while comprehensive abilities are required to handle tasks under various scenarios in the real world. In this paper, we propose a general Multi-Skill Dialog Framework, namely MSDF, which can be applied in different dialog tasks (e.g. knowledge grounded dialog and persona based dialog). Specifically, we propose a transferable response generator pre-trained on diverse large-scale dialog corpora as the backbone of MSDF, consisting of BERT-based encoders and a GPT-based decoder. To select the response consistent with dialog history, we propose a consistency selector trained through negative sampling. Moreover, the flexible copy mechanism of external knowledge is also employed to enhance the utilization of multiform knowledge in various scenarios. We conduct experiments on knowledge grounded dialog, recommendation dialog, and persona based dialog tasks. The experimental results indicate that our MSDF outperforms the baseline models with a large margin. In the Multi-skill Dialog of 2021 Language and Intelligence Challenge, our general MSDF won the 3rd prize, which proves our MSDF is effective and competitive.