Deep-speare: A Joint Neural Model of Poetic Language, Meter and Rhyme

  • 2018-07-10 06:26:24
  • Jey Han Lau, Trevor Cohn, Timothy Baldwin, Julian Brooke, Adam Hammond
  • 4

Abstract

In this paper, we propose a joint architecture that captures language, rhymeand meter for sonnet modelling. We assess the quality of generated poems usingcrowd and expert judgements. The stress and rhyme models perform very well, asgenerated poems are largely indistinguishable from human-written poems. Expertevaluation, however, reveals that a vanilla language model captures meterimplicitly, and that machine-generated poems still underperform in terms ofreadability and emotion. Our research shows the importance expert evaluationfor poetry generation, and that future research should look beyond rhyme/meterand focus on poetic language.

 

Quick Read (beta)

loading the full paper ...