Ernests Lavrinovics

Ernests Lavrinovics is a PhD fellow at the Department of Computer Science, Aalborg University, affiliated with the Technical Faculty of IT and Design and Data, Knowledge and Web Engineering research group. His work focuses on Natural Language Processing (NLP), Large Language Models (LLMs), and addressing challenges in machine translation and knowledge graph integration for resource-poor languages.

His research explores critical NLP themes including hallucination detection in LLMs, adapter-based regularization techniques for multilingual transfer, and multitask benchmark development for creole languages. He contributes to datasets like MultiHal and CreoleVal, emphasizing semantic grounding and cross-lingual evaluation.

• Recent publications analyze adapter architectures as regularizers in low-resource settings (ACL 2025)
• He co-developed MultiHal (2025), a multilingual dataset for LLM hallucination evaluation
• His work spans knowledge graph applications, transfer learning, and multilingual representation challenges

As an active academic, he has organized events like the Copenhagen NLP Symposium (2025) and AAU NLP Symposium (2024). He participates in technical programs at venues such as ACL and MRL Workshops, with citations in Scopus and Mendeley readership.