Approximating Interactive Human Evaluation with Self-Play for Open-Domain Dialog Systems

TL;DR

This paper introduces a novel self-play based evaluation method for open-domain dialog systems that better correlates with human judgment than existing automated metrics.

Contribution

It proposes a model-agnostic, dataset-agnostic self-play evaluation approach that approximates interactive human evaluation, improving the assessment of dialog quality.

Findings

Self-play metric correlates with human ratings (r>.7, p<.05).

The method outperforms existing automated metrics.

Open-sourced evaluation platform and dataset.

Abstract

Building an open-domain conversational agent is a challenging problem. Current evaluation methods, mostly post-hoc judgments of static conversation, do not capture conversation quality in a realistic interactive context. In this paper, we investigate interactive human evaluation and provide evidence for its necessity; we then introduce a novel, model-agnostic, and dataset-agnostic method to approximate it. In particular, we propose a self-play scenario where the dialog system talks to itself and we calculate a combination of proxies such as sentiment and semantic coherence on the conversation trajectory. We show that this metric is capable of capturing the human-rated quality of a dialog model better than any automated metric known to-date, achieving a significant Pearson correlation (r>.7, p<.05). To investigate the strengths of this novel metric and interactive evaluation in comparison to state-of-the-art metrics and human evaluation of static conversations, we perform extended experiments with a set of models, including several that make novel improvements to recent hierarchical dialog generation architectures through sentiment and semantic knowledge distillation on the utterance level. Finally, we open-source the interactive evaluation platform we built and the dataset we collected to allow researchers to efficiently deploy and evaluate dialog models.