SummEval: Re-evaluating Summarization Evaluation
This work addresses the problem of inconsistent evaluation protocols for researchers in text summarization, though it is incremental as it builds on existing datasets and metrics.
The authors tackled the lack of comprehensive evaluation protocols for text summarization by re-evaluating 14 automatic metrics and benchmarking 23 models using neural summarization outputs with human annotations, and they assembled the largest collection of model-generated summaries and human judgments for the CNN/DailyMail dataset.
The scarcity of comprehensive up-to-date studies on evaluation metrics for text summarization and the lack of consensus regarding evaluation protocols continue to inhibit progress. We address the existing shortcomings of summarization evaluation methods along five dimensions: 1) we re-evaluate 14 automatic evaluation metrics in a comprehensive and consistent fashion using neural summarization model outputs along with expert and crowd-sourced human annotations, 2) we consistently benchmark 23 recent summarization models using the aforementioned automatic evaluation metrics, 3) we assemble the largest collection of summaries generated by models trained on the CNN/DailyMail news dataset and share it in a unified format, 4) we implement and share a toolkit that provides an extensible and unified API for evaluating summarization models across a broad range of automatic metrics, 5) we assemble and share the largest and most diverse, in terms of model types, collection of human judgments of model-generated summaries on the CNN/Daily Mail dataset annotated by both expert judges and crowd-source workers. We hope that this work will help promote a more complete evaluation protocol for text summarization as well as advance research in developing evaluation metrics that better correlate with human judgments.