# SummPip: Unsupervised Multi-Document Summarization with Sentence Graph Compression

arXiv: https://arxiv.org/abs/2007.08954 · HTML: https://mlsanity.com/paper/2007.08954/

## Key facts

- Cited by 24 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 17 Jul 2020.

## Built on by (most cited first, 24 of 24)

- [An Empirical Survey on Long Document Summarization: Datasets, Models and Metrics](https://mlsanity.com/paper/2207.00939/) (2022)
- [Evaluating Open-QA Evaluation](https://mlsanity.com/paper/2305.12421/) (2023)
- [Peek Across: Improving Multi-Document Modeling via Cross-Document Question-Answering](https://mlsanity.com/paper/2305.15387/) (2023)
- [Target-aware Abstractive Related Work Generation with Contrastive Learning](https://mlsanity.com/paper/2205.13339/) (2022)
- [Leveraging Information Bottleneck for Scientific Document Summarization](https://mlsanity.com/paper/2110.01280/) (2021)
- [A Character-Level Length-Control Algorithm for Non-Autoregressive Sentence Summarization](https://mlsanity.com/paper/2205.14522/) (2022)
- [Template-based Abstractive Microblog Opinion Summarisation](https://mlsanity.com/paper/2208.04083/) (2022)
- [PDSum: Prototype-driven Continuous Summarization of Evolving Multi-document Sets Stream](https://mlsanity.com/paper/2302.05550/) (2023)
- [Precise Zero-Shot Pointwise Ranking with LLMs through Post-Aggregated Global Context Information](https://mlsanity.com/paper/2506.10859/) (2025)
- [Modeling Multi-level Context for Informational Bias Detection by Contrastive Learning and Sentential Graph Network](https://mlsanity.com/paper/2201.10376/) (2022)
- [Multi-Task Text Classification using Graph Convolutional Networks for Large-Scale Low Resource Language](https://mlsanity.com/paper/2205.01204/) (2022)
- [50 Ways to Bake a Cookie: Mapping the Landscape of Procedural Texts](https://mlsanity.com/paper/2210.17235/) (2022)
- [Benchmarking LLMs on the Semantic Overlap Summarization Task](https://mlsanity.com/paper/2402.17008/) (2024)
- [CNTLS: A Benchmark Dataset for Abstractive or Extractive Chinese Timeline Summarization](https://mlsanity.com/paper/2105.14201/) (2021)
- [Document-aware Positional Encoding and Linguistic-guided Encoding for Abstractive Multi-document Summarization](https://mlsanity.com/paper/2209.05929/) (2022)
- [Disentangling Specificity for Abstractive Multi-document Summarization](https://mlsanity.com/paper/2406.00005/) (2024)
- [GLIMMER: Incorporating Graph and Lexical Features in Unsupervised Multi-Document Summarization](https://mlsanity.com/paper/2408.10115/) (2024)
- [Multi-Narrative Semantic Overlap Task: Evaluation and Benchmark](https://mlsanity.com/paper/2201.05294/) (2022)
- [Large-Scale Multi-Document Summarization with Information Extraction and Compression](https://mlsanity.com/paper/2205.00548/) (2022)
- [GAE-ISumm: Unsupervised Graph-Based Summarization of Indian Languages](https://mlsanity.com/paper/2212.12937/) (2022)
- [Mining both Commonality and Specificity from Multiple Documents for Multi-Document Summarization](https://mlsanity.com/paper/2303.02677/) (2023)
- [MRGSEM-Sum: An Unsupervised Multi-document Summarization Framework based on Multi-Relational Graphs and Structural Entropy Minimization](https://mlsanity.com/paper/2507.23400/) (2025)
- [Large Language Model-Powered Query-Driven Event Timeline Summarization in Industrial Search](https://mlsanity.com/paper/2605.27066/) (2026)
- [SinBrief: A Hybrid Framework for Abstractive Text Summarisation of Sinhala Legal Documents](https://mlsanity.com/paper/2609.32397/) (2026)

## Abstract (arXiv)

Obtaining training data for multi-document summarization (MDS) is time consuming and resource-intensive, so recent neural models can only be trained for limited domains. In this paper, we propose SummPip: an unsupervised method for multi-document summarization, in which we convert the original documents to a sentence graph, taking both linguistic and deep representation into account, then apply spectral clustering to obtain multiple clusters of sentences, and finally compress each cluster to generate the final summary. Experiments on Multi-News and DUC-2004 datasets show that our method is competitive to previous unsupervised methods and is even comparable to the neural supervised approaches. In addition, human evaluation shows our system produces consistent and complete summaries compared to human written ones.