Enhancing Abstractiveness of Summarization Models through Calibrated Distillation
This work addresses a specific bottleneck in making summarization models more efficient and abstractive, but it is incremental as it builds on existing distillation techniques.
The paper tackled the problem of reduced abstractiveness in sequence-level knowledge distillation for abstractive summarization by proposing DisCal, which uses diverse pseudo summaries and dual supervision to improve abstractiveness (measured by n-gram overlap) while maintaining informativeness (measured by ROUGE). The method outperformed prior approaches in experiments.
Sequence-level knowledge distillation reduces the size of Seq2Seq models for more efficient abstractive summarization. However, it often leads to a loss of abstractiveness in summarization. In this paper, we propose a novel approach named DisCal to enhance the level of abstractiveness (measured by n-gram overlap) without sacrificing the informativeness (measured by ROUGE) of generated summaries. DisCal exposes diverse pseudo summaries with two supervision to the student model. Firstly, the best pseudo summary is identified in terms of abstractiveness and informativeness and used for sequence-level distillation. Secondly, their ranks are used to ensure the student model to assign higher prediction scores to summaries with higher ranks. Our experiments show that DisCal outperforms prior methods in abstractive summarization distillation, producing highly abstractive and informative summaries.