Institutional Repository of Vilnius University
Didelių kalbos modelių (DKM) generuoto edukacinio turinio vertinimas /
Abstract
dc:descriptionThis thesis explores automatic question generation and answer evaluation for Lithuanian and English texts using large language models (LLMs). During the study, a dataset containing Lithuanian and English texts of comparable sizes was compiled, from which questionnaires were generated. There were two question generation and answer evaluation methods defined, termed the simple and the complex method, and implemented with two large language models, OpenAI's o3-mini and Alibaba's Qwen2.5-Max - creating four distinct methodologies. Results showed that in general, the methodologies performed better in English than in Lithuanian. It was found that the proposed complex method underperformed or showed similar results compared to the simpler method, which had several times lower operational costs. Overall, the most effective methodology was the simple method using OpenAI's o3-mini large language model. Finally, the thesis outlines future research directions with the themes of question quality, automatic verification of question correctness, and analysis of patterns in misclassified answers.
Degree
thesis:*- Grantor dc:publisher
- Institutional Repository of Vilnius University
- Year dc:date
- 2025
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Kilinskas, Rokas,
Rights
dc:rights- Statement dc:rights
-
- info:eu-repo/semantics/openAccess
- Language dc:language
- lit
Identifiers
dc:identifier.*- Repository record dc:identifier
- https://repository.vu.lt/VU:ELABAETD239031085&prefLang=en_US