How Far are We from Robust Long Abstractive Summarization?
This work addresses the problem of ensuring factual accuracy in long abstractive summaries for researchers and practitioners, but it is incremental as it builds on existing evaluation methods.
The study evaluated long document abstractive summarization systems through fine-grained human annotations, finding that optimizing for ROUGE scores improves relevance but not factuality, and ROUGE remains best for relevance while factuality metrics have limitations.
Abstractive summarization has made tremendous progress in recent years. In this work, we perform fine-grained human annotations to evaluate long document abstractive summarization systems (i.e., models and metrics) with the aim of implementing them to generate reliable summaries. For long document abstractive models, we show that the constant strive for state-of-the-art ROUGE results can lead us to generate more relevant summaries but not factual ones. For long document evaluation metrics, human evaluation results show that ROUGE remains the best at evaluating the relevancy of a summary. It also reveals important limitations of factuality metrics in detecting different types of factual errors and the reasons behind the effectiveness of BARTScore. We then suggest promising directions in the endeavor of developing factual consistency metrics. Finally, we release our annotated long document dataset with the hope that it can contribute to the development of metrics across a broader range of summarization settings.