Year: 2024
Type: dissertation
Abstract: The object of development is a code based on large language models for generating dramatic texts. The object of the study is a dramatic text generated by a large language model of machine learning. The subject of the study is methods based on large language models for generating dramatic text. The purpose of the work is to create a generator of dramatic text, taking into account the features of the dramatic style of narration, without the direct participation of the author. The work considers the features of the dramatic text, as well as the problems that arise during its generation. To solve them, the method of hierarchical plot generation was used, with explicit narrative structures and characters, which helps to generate more coherent stories, especially when creating long texts such as theater scripts. To preserve the style, a set of written prefixes taken from the ancient Greek tragedy "Medea" by Euripides (431 BC) was used. The evaluation was carried out using the NLI-score metric and by surveying people involved in script writing. An algorithm for generating and evaluating scripts was written in the Python programming language.
Authors: Данилов, Е. М., Danilov, E. M.
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7019215598