How NOT To Evaluate Your Dialogue System: An Empirical Study of Unsupervised Evaluation Metrics for Dialogue Response Generation

We investigate evaluation metrics for endto-end dialogue systems where supervised labels, such as task completion, are not available. Recent works in end-to-end dialogue systems have adopted metrics from machine translation and text summarization to compare a model’s generated response to a single target response. We show that these metrics correlate very weakly or not at all with human judgements of the response quality in both technical and non-technical domains. We provide quantitative and qualitative results highlighting specific weaknesses in existing metrics, and provide recommendations for future development of better automatic evaluation metrics for dialogue systems.