Q^2: Evaluating Factual Consistency in Knowledge-Grounded Dialogues via Question Generation and Question Answering
Abstract
Neural knowledge-grounded generative models for dialogue often produce content that is \textit{factually inconsistent} with the knowledge they rely on, making them unreliable and limiting their applicability.
Inspired by recent work on evaluating factual consistency in abstractive summarization, we propose an automatic evaluation metric for factual consistency in knowledge-grounded dialogue using automatic question generation and question answering. Our metric, denoted $Q^2$, compares answer spans using natural language inference, which enables better factual comparison than in previous token-based metrics. To foster proper evaluation, we curate a novel dataset of state-of-the-art dialogue system outputs for the Wizard-of-Wikipedia dataset, manually annotated for factual consistency. We perform a thorough meta-evaluation of $Q^2$ against other metrics using the new dataset and two others, where it shows higher correlation with human judgements.
Inspired by recent work on evaluating factual consistency in abstractive summarization, we propose an automatic evaluation metric for factual consistency in knowledge-grounded dialogue using automatic question generation and question answering. Our metric, denoted $Q^2$, compares answer spans using natural language inference, which enables better factual comparison than in previous token-based metrics. To foster proper evaluation, we curate a novel dataset of state-of-the-art dialogue system outputs for the Wizard-of-Wikipedia dataset, manually annotated for factual consistency. We perform a thorough meta-evaluation of $Q^2$ against other metrics using the new dataset and two others, where it shows higher correlation with human judgements.
Research Areas
