USR: An Unsupervised and Reference Free Evaluation Metric for Dialog Generation
This addresses the lack of effective automatic evaluation metrics for open-domain dialog research, which has been a bottleneck in the field.
The paper tackles the problem of evaluating dialog generation by proposing USR, an unsupervised and reference-free metric that strongly correlates with human judgment, achieving turn-level correlations of 0.42-0.48 and system-level correlations of 1.0 on datasets like Topical-Chat and PersonaChat.
The lack of meaningful automatic evaluation metrics for dialog has impeded open-domain dialog research. Standard language generation metrics have been shown to be ineffective for evaluating dialog models. To this end, this paper presents USR, an UnSupervised and Reference-free evaluation metric for dialog. USR is a reference-free metric that trains unsupervised models to measure several desirable qualities of dialog. USR is shown to strongly correlate with human judgment on both Topical-Chat (turn-level: 0.42, system-level: 1.0) and PersonaChat (turn-level: 0.48 and system-level: 1.0). USR additionally produces interpretable measures for several desirable properties of dialog.