Thesis (Selection of subject)Thesis (Selection of subject)(version: 368)
Thesis details
   Login via CAS
Converting prose into poetry using neural networks
Thesis title in Czech: Převod prózy do poezie pomocí neuronových sítí
Thesis title in English: Converting prose into poetry using neural networks
Key words: generování poezie, strojový překlad, hluboké neuronové sítě, Transformer
English key words: generating poetry, machine translation, deep neural networks, Transformer
Academic year of topic announcement: 2019/2020
Thesis type: diploma thesis
Thesis language: angličtina
Department: Institute of Formal and Applied Linguistics (32-UFAL)
Supervisor: Mgr. Martin Popel, Ph.D.
Author: hidden - assigned and confirmed by the Study Dept.
Date of registration: 03.01.2020
Date of assignment: 03.01.2020
Confirmed by Study dept. on: 13.01.2020
Date and time of defence: 08.09.2021 09:00
Date of electronic submission:23.07.2021
Date of submission of printed version:23.07.2021
Date of proceeded defence: 08.09.2021
Opponents: Mgr. et Mgr. Ondřej Dušek, Ph.D.
This thesis investigates the applications of sequence to sequence neural network models in the generation of poetry, particularly harnessing machine translation methods to capture the difference between prose and poetry, and "translate" prose into poems. Style, meaning, rhyme, rhythm and structure are some of the properties that the investigation will focus on. Collections of Czech, English and Turkish poetry and prose will serve as training data, and will be used along with machine translation models to produce prose versions of the poems as well as training new MT models to translate the generated prose into poetry.
Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N Gomez, Łukasz Kaiser, and Illia Polosukhin. Attention is All you Need. In Guyon, I., U. V. Luxburg, S. Bengio, H. Wallach, R. Fergus, S. Vishwanathan, and R. Garnett, editors, Advances in Neural Information Processing Systems 30, pages 6000–6010. Curran Associates, Inc., 2017. URL

Sennrich, Rico, Barry Haddow, and Alexandra Birch. Improving neural machine translation models with monolingual data. ACL 2016.

Bei Liu Jianlong Fu Makoto P. Kato Masatoshi Yoshikawa. Beyond Narrative Description: Generating Poetry from Images by Multi-Adversarial Training. ACM Multimedia 2018.
Preliminary scope of work in English
There are many approaches for automatic generation of verses using (neural or n-gram) language models - the user then writes first few words and the machine completes the poem.
However, the goal of this thesis is different - the user inputs a text in prose and the machine converts it into verses, while preserving the meaning to some extent.
Part of the thesis will be designing and implementing a set of automatic evaluation measures.
Charles University | Information system of Charles University |