Multi-Sentence Compression with Word Vertex-Labeled Graphs and Integer Linear Programming

Elvys Linhares Pontes 1 Stéphane Huet 1 Thiago Gouveia da Silva 2 Juan-Manuel Torres-Moreno 1, 3 Andréa Carneiro Linhares 4
1 LIA - Laboratoire Informatique d'Avignon
2 Inst. Federal de Educação, Ciência e Tecnologia da Paraı́ba (IFPB)
3 EPM - École Polytechnique de Montréal
4 UFC - Universidade Federal do Ceará
Abstract : Multi-Sentence Compression (MSC) aims to generate a short sentence with key information from a cluster of closely related sentences. MSC enables summarization and question-answering systems to generate outputs combining fully formed sentences from one or several documents. This paper describes a new Integer Linear Programming method for MSC using a vertex-labeled graph to select different keywords, and novel 3-grams scores to generate more informative sentences while maintaining their grammaticality. Our system is of good quality and outperforms the state-of-the-art for evaluations led on news dataset. We led both automatic and manual evaluations to determine the informativeness and the gram-maticality of compressions for each dataset. Additional tests, which take advantage of the fact that the length of compressions can be modulated, still improve ROUGE scores with shorter output sentences.
Document type :
Conference papers
Complete list of metadatas

Cited literature [22 references]  Display  Hide  Download
https://hal.archives-ouvertes.fr/hal-01779470
Contributor : Elvys Linhares Pontes <>
Submitted on : Thursday, April 26, 2018 - 3:54:38 PM
Last modification on : Friday, March 22, 2019 - 11:34:07 AM
Long-term archiving on : Tuesday, September 25, 2018 - 12:29:07 PM

File

TextGraphs_2018.pdf
Files produced by the author(s)

Identifiers

Collections

UNIV-AVIGNON

Citation

Elvys Linhares Pontes, Stéphane Huet, Thiago Gouveia da Silva, Juan-Manuel Torres-Moreno, Andréa Carneiro Linhares. Multi-Sentence Compression with Word Vertex-Labeled Graphs and Integer Linear Programming. NAACL-HLT 12th Workshop on Graph-based Methods for Natural Language Processing (TextGraphs), 2018, New Orleans, LA, United States. ⟨10.18653/v1/W18-1704⟩. ⟨hal-01779470⟩

Export

BibTeX TEI DC DCterms EndNote

Share

Metrics

Record views

62

Files downloads

124

Contact

support.ccsd.cnrs.fr
hal.support@ccsd.cnrs.fr
Logo CCSD