Speech Sentiment Classification in a Ukrainian-Russian Environment

dc.contributor.author	Patenko, Liudmyla
dc.date.accessioned	2024-08-23T09:11:58Z
dc.date.available	2024-08-23T09:11:58Z
dc.date.issued	2024
dc.identifier.citation	Patenko Liudmyla. Speech Sentiment Classification in a Ukrainian-Russian Environment. Ukrainian Catholic University, Faculty of Applied Sciences, Department of Computer Sciences. Lviv 2024, 39 p.	uk
dc.identifier.uri	https://er.ucu.edu.ua/handle/1/4675
dc.language.iso	en	uk
dc.subject	Speech Sentiment Classification	uk
dc.subject	Ukrainian-Russian Environment	uk
dc.title	Speech Sentiment Classification in a Ukrainian-Russian Environment	uk
dc.type	Preprint	uk
dc.status	Публікується вперше	uk
dc.description.abstracten	The process of sentiment classification involves categorizing human speech into one or more classes based on the emotional information expressed by the speakers. This study is focused on the development of a Speech Sentiment Classification (SSC) sys- tem designed to classify sentiment in a multi-lingual environment, including the Ukrainian language, while addressing the challenge of data scarcity. The research presents and evaluates three distinct approaches to this problem: a text-only clas- sifier utilizing a Large Language Model (LLM), an audio-only classifier, and a bi- modal fusion approach that combines both text and audio features. The results indi- cate that the bi-modal fusion approach achieved an accuracy of 85% and an F1 score of 0.85 for binary classification of negative versus neutral sentiment.	uk