Guess who? Multilingual approach for the automated generation of author-stylized poetry

ABSTRACT This paper addresses the problem of stylized text generation in a multilingual setup. A version of a language model based on a long short-term memory (LSTM) artificial neural network with extended phonetic and semantic embeddings is used for stylized poetry generation. Phonetics is shown to have comparable importance for the task of stylized poetry generation as the information on the target author. The quality of the resulting poems generated by the network is estimated through bilingual evaluation understudy (BLEU), a survey and a new cross-entropy based metric that is suggested for the problems of such type. The experiments show that the proposed model consistently outperforms random sample and vanilla-LSTM baselines, humans also tend to attribute machine generated texts to the target author. Index Terms-- stylized text generation, poetry generation, artificial neural networks, multilingual models 1. INTRODUCTION The problem of making machine-generated text feel more authentic has a number of industrial and scientific applications, see, for example, [1] or [2]. Most modern generative models are trained on huge corpora of texts which include different contributions from various authors.