Unifying Human and Statistical Evaluation for Natural Language Generation
This addresses the evaluation bottleneck for NLG systems, offering a more comprehensive metric for researchers and practitioners, though it is incremental as it builds on existing human and statistical methods.
The paper tackles the problem of evaluating both quality and diversity in natural language generation systems by proposing HUSE, a unified framework based on the optimal error rate of distinguishing human- from machine-generated text, and demonstrates that it detects diversity defects in summarization and chit-chat dialogue where human evaluation fails.
How can we measure whether a natural language generation system produces both high quality and diverse outputs? Human evaluation captures quality but not diversity, as it does not catch models that simply plagiarize from the training set. On the other hand, statistical evaluation (i.e., perplexity) captures diversity but not quality, as models that occasionally emit low quality samples would be insufficiently penalized. In this paper, we propose a unified framework which evaluates both diversity and quality, based on the optimal error rate of predicting whether a sentence is human- or machine-generated. We demonstrate that this error rate can be efficiently estimated by combining human and statistical evaluation, using an evaluation metric which we call HUSE. On summarization and chit-chat dialogue, we show that (i) HUSE detects diversity defects which fool pure human evaluation and that (ii) techniques such as annealing for improving quality actually decrease HUSE due to decreased diversity.