Planned intervention: On Wednesday April 3rd 05:30 UTC Zenodo will be unavailable for up to 2-10 minutes to perform a storage cluster upgrade.
Published September 9, 2017 | Version v1
Conference paper Open

Split and Rephrase

  • 1. University of Edinburgh
  • 2. CNRS

Description

We propose a new sentence simplification task (Split-and-Rephrase) where the aim is to split a complex sentence into a meaning preserving sequence of shorter sentences. Like sentence simplification, splitting-and-rephrasing has the potential of benefiting both natural language processing and societal applications. Because shorter sentences are generally better processed by NLP systems, it could be used as a preprocessing step which facilitates and improves the performance of parsers, semantic role labellers and machine translation systems. It should also be of use for people with reading disabilities because it allows the conversion of longer
sentences into shorter ones. This paper makes two contributions towards this new task. First, we create and make available
a benchmark consisting of 1,066,115 tuples mapping a single complex sentence to a sequence of sentences expressing the same meaning. Second, we propose five models (vanilla sequence-to-sequence to semantically-motivated models) to understand the difficulty of the proposed task.

Files

2017-EMNLP-simplification.arxiv.pdf

Files (264.0 kB)

Name Size Download all
md5:901ca8dd2b05d983506f977618712e0a
264.0 kB Preview Download

Additional details

Funding

SUMMA – Scalable Understanding of Multilingual Media 688139
European Commission