Comparative Study of Transformer Models for Emotion Classification in User-Generated Text

Student: Matvei Kirdin

Educational Programme: Data Science(Master)

Year of Graduation: 2026

An ability to understand emotions expressed through text using a machine is crucial in many business, political, and scientific applications. Research in this area is twofold with some experimenting with novel model architectures and others providing large-scale labeled datasets. This work introduces a dataset of manually labeled Bluesky user posts according to Ekhart emotion typology consisting of 6 emotions such as anger, disgust, joy, sadness, surprise, and fear along with neutral category. Several state of the art transformer models were evaluated on it in different scenarios such as fine-tuning and transfer learning. Fine-tuned embedding models show the best results in terms of valuation metrics, however they lack the generalisability of generative models.

Student Theses at HSE must be completed in accordance with the University Rules and regulations specified by each educational programme.

Summaries of all theses must be published and made freely available on the HSE website.

The full text of a thesis can be published in open access on the HSE website only if the authoring student (copyright holder) agrees, or, if the thesis was written by a team of students, if all the co-authors (copyright holders) agree. After a thesis is published on the HSE website, it obtains the status of an online publication.

Student theses are objects of copyright and their use is subject to limitations in accordance with the Russian Federation’s law on intellectual property.

In the event that a thesis is quoted or otherwise used, reference to the author’s name and the source of quotation is required.

Search all student theses