The Glass Ceiling of Automatic Evaluation in Natural Language Generation
This work reveals fundamental limitations in current automatic evaluation methods for NLG, which could hinder progress in the field if not addressed.
The paper analyzes automatic evaluation metrics for natural language generation by comparing their system rankings with human judgments, finding that automatic metrics are much more similar to each other than to humans and fail to complement each other effectively.
Automatic evaluation metrics capable of replacing human judgments are critical to allowing fast development of new methods. Thus, numerous research efforts have focused on crafting such metrics. In this work, we take a step back and analyze recent progress by comparing the body of existing automatic metrics and human metrics altogether. As metrics are used based on how they rank systems, we compare metrics in the space of system rankings. Our extensive statistical analysis reveals surprising findings: automatic metrics -- old and new -- are much more similar to each other than to humans. Automatic metrics are not complementary and rank systems similarly. Strikingly, human metrics predict each other much better than the combination of all automatic metrics used to predict a human metric. It is surprising because human metrics are often designed to be independent, to capture different aspects of quality, e.g. content fidelity or readability. We provide a discussion of these findings and recommendations for future work in the field of evaluation.