PECULIARITIESS OF LITERARY GENRE CLASSIFICATION USING LARGE LANGUAGE MODELS

Abstract

The article examines the historical context of genre classification, the parameterization of genres, and their role in structuring literary works. The experience of applying automated parameterization in the fields of films (Netflix) and music (Spotify, Apple Music) is analyzed, demonstrating how it enables personalized user experiences. Similar methods are used in the literary sector, particularly on Amazon, where genre parameterization aids in book sales and recommendations. In scientific literature, genre classification also plays a key role in knowledge systematization, contributing to more efficient search and analysis of academic works.One of the central issues addressed in the article is the role of large language models in genre classification. The study explores how these models can identify genre-specific features of texts by highlighting stylistic techniques and considering lexical and syntactic parameters. Particular attention is given to methods for assessing the accuracy of automated classification, as well as analyzing confusion matrices and evaluating model confidence. It was emphasized that integrating large language models with genre ontologies and refining adaptive algorithms could significantly enhance the quality of automated classification.The article also compares traditional and automated approaches to genre classification. Traditional methods rely on expert analysis and take into account the cultural and historical context of genres, providing deeper analytical insights but limiting scalability. Automated methods, on the other hand, allow for the rapid processing of large volumes of text but may overlook nuances and genre hybridity. The optimal approach is a combination of these methods, ensuring a balance between accuracy and interpretability.Among the key challenges in genre classification using large language models, the study highlights the ambiguity of genre boundaries, the impact of training data biases on model performance, and the difficulty of recognizing complex stylistic techniques. Suggested improvements include expanding training data corpora, integrating models with literary knowledge bases, implementing multi-level classification strategies, and enhancing the explainability of classification results.Thus, the article makes a significant contribution to research on automated genre classification, demonstrating both the advantages and limitations of large language models in this task.