| Title |
Mašininio mokymo taikymo klausimų kūrimui bei atsakymui lietuvių kalba tyrimas |
| Translation of Title |
A study of the application of machine learning to answering and generating questions in lithuanian. |
| Authors |
Talačka, Kristupas |
| Full Text |
|
| Pages |
57 |
| Keywords [eng] |
machine learning ; natural language processing ; text-generating artificial intelligence models |
| Abstract [eng] |
This project presents a system for automatic question generation and answering using natural language processing (NLP) methods in the Lithuanian language. The system is designed to support educational and business processes, allowing users to generate questions and answers using either pre-trained or custom fine-tuned artificial intelligence models. The literature review analyzes existing methods for question generation and answering, their application to low-resource languages such as Lithuanian, and compares model architectures, datasets, and evaluation metrics. The experimental part describes datasets created using automated data augmentation methods and presents training and evaluation experiments. The experiments compare models trained solely on the original data and models trained using augmented datasets. The results show that augmentation methods significantly improve question answering and question generation model performance. The best results were achieved using the back-translation technique. |
| Dissertation Institution |
Kauno technologijos universitetas. |
| Type |
Master thesis |
| Language |
Lithuanian |
| Publication date |
2025 |
