We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

cs.CL

Change to browse by:

cs

References & Citations

DBLP - CS Bibliography

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo

Computer Science > Computation and Language

Title: Unsupervised Summarization by Jointly Extracting Sentences and Keywords

Abstract: We present RepRank, an unsupervised graph-based ranking model for extractive multi-document summarization in which the similarity between words, sentences, and word-to-sentence can be estimated by the distances between their vector representations in a unified vector space. In order to obtain desirable representations, we propose a self-attention based learning method that represent a sentence by the weighted sum of its word embeddings, and the weights are concentrated to those words hopefully better reflecting the content of a document. We show that salient sentences and keywords can be extracted in a joint and mutual reinforcement process using our learned representations, and prove that this process always converges to a unique solution leading to improvement in performance. A variant of absorbing random walk and the corresponding sampling-based algorithm are also described to avoid redundancy and increase diversity in the summaries. Experiment results with multiple benchmark datasets show that RepRank achieved the best or comparable performance in ROUGE.
Comments: 10 pages(includes 2 pages references), 1 figure
Subjects: Computation and Language (cs.CL)
Cite as: arXiv:2009.07481 [cs.CL]
  (or arXiv:2009.07481v2 [cs.CL] for this version)

Submission history

From: Zongyi Li [view email]
[v1] Wed, 16 Sep 2020 05:58:00 GMT (60kb,D)
[v2] Mon, 24 Jul 2023 03:26:17 GMT (60kb,D)

Link back to: arXiv, form interface, contact.