Self-Supervised Learning for Contextualized Extractive Summarization
This work addresses the need for better contextual understanding in extractive summarization, offering a novel pre-training approach that improves performance for NLP applications.
The paper tackled the problem of extractive summarization lacking explicit document-level context by introducing three auxiliary self-supervised pre-training tasks, resulting in a simple model that outperforms previous state-of-the-art methods on the CNN/DM dataset.
Existing models for extractive summarization are usually trained from scratch with a cross-entropy loss, which does not explicitly capture the global context at the document level. In this paper, we aim to improve this task by introducing three auxiliary pre-training tasks that learn to capture the document-level context in a self-supervised fashion. Experiments on the widely-used CNN/DM dataset validate the effectiveness of the proposed auxiliary tasks. Furthermore, we show that after pre-training, a clean model with simple building blocks is able to outperform previous state-of-the-art that are carefully designed.