Automatic Poetry Generation from Prosaic Text

Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pages 2471–2480 July 5 - 10, 2020. c⃝2020 Association for Computational Linguistics 2471 Automatic Poetry Generation from Prosaic Text Tim Van de Cruys Institut de Recherche en Informatique de Toulouse (IRIT) Artificial and Natural Intelligence Toulouse Institute (ANITI) CNRS, Toulouse tim.vandecruys@irit.fr Abstract In the last few years, a number of successful approaches have emerged that are able to ad- equately model various aspects of natural lan- guage. In particular, language models based on neural networks have improved the state of the art with regard to predictive language mod- eling, while topic models are successful at cap- turing clear-cut, semantic dimensions. In this paper, we explore how these approaches can be adapted and combined to model the linguis- tic and literary aspects needed for poetry gen- eration. The system is exclusively trained on standard, non-poetic text, and its output is con- strained in order to confer a poetic character to the generated verse. The framework is applied to the generation of poems in both English and French, and is equally evaluated for both lan- guages. Even though it only uses standard, non-poetic text as input, the system yields state of the art results for poetry generation. 1 Introduction Automatic poetry generation is a challenging task for a computational system. For a poem to be mean- ingful, both linguistic and literary aspects need to be taken into account. First of all, a poetry gen- eration system needs to properly model language phenomena, such as syntactic well-formedness and topical coherence. Furthermore, the system needs to incorporate various constraints (such as form and rhyme) that are related to a particular poetic genre. And finally, the system needs to exhibit a certain amount of literary creativity, which makes the poem interesting and worthwhile to read. In recent years, a number of fruitful NLP ap- proaches have emerged that are able to adequately model various aspects of natural language. In par- ticular, neural network language models have im- proved the state of the art in language modeling, while topic models are successful at capturing clear- cut, semantic dimensions. In this paper, we explore how these approaches can be adapted and com- bined in order to model both the linguistic and literary aspects that are required for poetry genera- tion. More specifically, we make use of recurrent neural networks in an encoder-decoder configura- tion. The encoder first constructs a representation of an entire sentence by sequentially incorporating each word of the sentence into a fixed-size hidden state vector. The final representation is then given to the decoder, which emits a sequence of words according to a probability distribution derived from the hidden state of the input sentence. By train- ing the network to predict the next sentence with the current sentence as input, the network learns to generate plain text with a certain discourse co- herence. By modifying the probability distribution yielded by the decoder, we enforce the incorpora- tion of poetic constraints, such that …

Content truncated to 500 words