Combining Graph Degeneracy and Submodularity for Unsupervised Extractive Summarization

Abstract

We present a fully unsupervised, extractive text summarization system that leverages a submodularity framework introduced by past research. The framework allows summaries to be generated in a greedy way while preserving near-optimal performance guarantees. Our main contribution is the novel coverage reward term of the objective function optimized by the greedy algorithm. This component builds on the graph-of-words representation of text and the k-core decomposition algorithm to assign meaningful scores to words. We evaluate our approach on the AMI and ICSI meeting speech corpora, and on the DUC2001 news corpus. We reach state-of-the-art performance on all datasets. Results indicate that our method is particularly well-suited to the meeting domain.

Discourse-Aware Unsupervised Summarization for Long Scientific Documents

“…Traditional extractive summarization methods are mostly unsupervised (Radev et al, 2000;Lin and Hovy, 2002;Wan, 2008;Wan and Yang, 2008;Hirao et al, 2013;Parveen et al, 2015;Yin and Pei, 2015;Li et al, 2017;Zheng and Lapata, 2019), utilizing a notion of sentence importance based on n-gram overlap with other sentences and frequency information (Nenkova and Vanderwende, 2005), relying on graph-based methods for sentence ranking (Erkan and Radev, 2004;Mihalcea and Tarau, 2004), or performing keyword extraction combined with submodular maximization (Tixier et al, 2017;Shang et al, 2018).…”

Improving Unsupervised Extractive Summarization with Facet-Aware Modeling

“…"..." refers to the omissions of context sentences due to space limitation. extractive summarization (Radev et al, 2000;Mihalcea and Tarau, 2004;Erkan and Radev, 2004;Schluter and Søgaard, 2015;Tixier et al, 2017;Zheng and Lapata, 2019;Xu et al, 2020;Dong et al, 2020). Compare with supervised ones, unsupervised methods 1).…”

Extractive Summarization of Long Documents by Combining Global and Local Context

“…The two datasets annotated with extractive labels will be made public. 1 2 Related work 2.1 Extractive summarization Traditional extractive summarization methods are mostly based on explicit surface features , relying on graph-based methods (Mihalcea and Tarau, 2004), or on submodular maximization (Tixier et al, 2017). Benefiting from the success of neural sequence models in other NLP tasks, Cheng and Lapata (2016) propose a novel approach to extractive summarization based on neural networks and continuous sentence features, which outperforms traditional methods on the Dai-lyMail dataset.…”