Research

My research focuses on the mathematical reasoning capabilities of language models. I study how changes in problem representation affect model performance, conduct human studies to compare reasoning patterns, and analyse how problem domain and complexity shape outcomes. I also develop methods for automated evaluation and synthetic data generation. My list of publications is available on Google Scholar 🎓

Publications

Neural Network Methods for Selecting and Generating Synthetic Variations of Combinatorial Problems.
Nikolaiev, Andrii, and Anisimov, Anatoliy.
Cybernetics and Systems Analysis, pp. 354–363. Springer Nature, 2025.
Can language models rival mathematics students? Evaluating mathematical reasoning through textual manipulation and human experiments.
Nikolaiev, Andrii, Stathopoulos, Yiannos, and Teufel, Simone.
ACL RR. arXiv pre-print, 2024.
Comparison of Problem-solving Performance Across Mathematical Domains with Large Language Models.
Nikolaiev, Andrii D., and Derevianchenko, Oleksandr V.
Artificial Intelligence Scientific Journal, pp. 96–104. 2024.

Scientific conferences

Introducing Constraints in Combinatorial Problems: A Case Study with LLaMA 3.1.
11th International Scientific Conference on Information Technology and Implementation (IT&I-2024), Kyiv, Ukraine, Dec. 2024.
AI in Education: Application of LLMs for Learning Mathematics.
Ukrainian Cambridge: New Research by Displaced Scholars from Ukraine, Cambridge, UK, Nov. 2023.
Mathematical Word Problem Solution Evaluation via Data Preprocessing Approach.
8th International Scientific Conference on Information Technology and Implementation (IT&I-2021), Kyiv, Ukraine, Dec. 2021.
Implementation of Artificial Intelligence Module for Educational Purposes.
7th International Scientific Conference on Information Technology and Interactions (IT&I-2020), Kyiv, Ukraine, Nov. 2020.