Research on the application of artificial intelligence in understanding, generating, and teaching language. Our research encompasses three interconnected areas:
AI in Education and Language Pedagogy. We study the impact of artificial intelligence on teaching with special focus on foreign language instruction. We investigate how AI has influenced glotodidactics (language teaching methodology) and how computer science students use AI tools in their education. We also work on predictive language analytics aiming to forecast various future trends in language and education.
Large Language Models for Croatian and Low-Resource Languages. We develop methodologies for domain adaptation of language models (fine-tuning, prompt engineering) and quantization/optimization techniques for local deployment with sensitive data. Special emphasis on evaluation frameworks for Croatian language and comparisons with other languages.
Machine Translation and Knowledge-Based Systems. We research domain-specific machine translation for specialized fields such as agriculture, and retrieval-augmented generation (RAG) systems combining language models with external knowledge bases.
Technologies Covered
- AI in education and e-learning
- Predictive language analytics
- Computer-assisted language learning (CALL)
- Large Language Models (LLaMA, Mistral, GPT)
- Fine-tuning and prompt engineering
- Model quantization and local deployment
- Retrieval-Augmented Generation (RAG)
- Neural Machine Translation (Tower, NLLB)
- Low-resource language processing
