We show that a simple unsupervised masking objective can approach near supervised performance on abstractive multi-document news summarization. Our method trains a state-of-the-art neural summarization model to predict the masked out source document with highest lexical centrality relative to the multi-document group. In experiments on the Multi-News dataset, our masked training objective yields a system that outperforms past unsupervised methods and, in human evaluation, surpasses the best supervised method without requiring access to any ground-truth summaries. Further, we evaluate how different measures of lexical centrality, inspired by past work on extractive summarization, affect final performance.
@article{arxiv.2201.02321,
title = {An Unsupervised Masking Objective for Abstractive Multi-Document News Summarization},
author = {Nikolai Vogler and Songlin Li and Yujie Xu and Yujian Mi and Taylor Berg-Kirkpatrick},
journal= {arXiv preprint arXiv:2201.02321},
year = {2022}
}