AdaptSum: Towards Low-Resource Domain Adaptation for Abstractive Summarization
This addresses the problem of limited generalization in summarization models for domains with scarce labeled data, though it is incremental as it builds on existing pre-training methods.
The paper tackled domain adaptation for abstractive summarization in low-resource settings by investigating pre-training strategies, finding that effectiveness correlates with data similarity and catastrophic forgetting can be mitigated, but a significant performance gap remains compared to high-resource settings.
State-of-the-art abstractive summarization models generally rely on extensive labeled data, which lowers their generalization ability on domains where such data are not available. In this paper, we present a study of domain adaptation for the abstractive summarization task across six diverse target domains in a low-resource setting. Specifically, we investigate the second phase of pre-training on large-scale generative models under three different settings: 1) source domain pre-training; 2) domain-adaptive pre-training; and 3) task-adaptive pre-training. Experiments show that the effectiveness of pre-training is correlated with the similarity between the pre-training data and the target domain task. Moreover, we find that continuing pre-training could lead to the pre-trained model's catastrophic forgetting, and a learning method with less forgetting can alleviate this issue. Furthermore, results illustrate that a huge gap still exists between the low-resource and high-resource settings, which highlights the need for more advanced domain adaptation methods for the abstractive summarization task.