Keywords: natural language processing, natural language generation, benchmark construction, evaluation
TL;DR: We develop a framework for generating controlled perturbations and identify subsets in text-to-scalar, text-to-text, or data-to-text settings, and release an evaluation suite that we test with state-of-the-art NLG models.
Abstract: Machine learning approaches applied to NLP are often evaluated by summarizing their performance in a single number, for example accuracy. Since most test sets are constructed as an i.i.d. sample from the overall data, this approach overly simplifies the complexity of language and encourages overfitting to the head of the data distribution. As such, rare language phenomena or text about underrepresented groups are not equally included in the evaluation. To encourage more in-depth model analyses, researchers have proposed the use of multiple test sets, also called challenge sets, that assess specific capabilities of a model. In this paper, we develop a framework based on this idea which is able to generate controlled perturbations and identify subsets in text-to-scalar, text-to-text, or data-to-text settings. By applying this framework to the GEM generation benchmark, we propose an evaluation suite made of 80 challenge sets, demonstrate the kinds of analyses that it enables and shed light onto the limits of current generation models.
Supplementary Material: zip
URL: https://gem-benchmark.com/data_cards (The datasets of our evaluation suite are described in the "Changes to the Original Dataset for GEM" section of the data card of the concerned datasets).
Contribution Process Agreement: Yes
Dataset Url: https://gem-benchmark.com/data_cards
Author Statement: Yes