Archaeology at BEA 2025 Shared Task: Are Simple Baselines Good Enough?

This paper describes our approach to the 5 classification tasks from the Building Educational Applications (BEA) 2025 Shared Task.Our methods range from classical machine learning models to fine-tuning large-scale transformer architectures.Despite the diversity of techniques, performance differences were often minor, suggesting the presence of strong surfacelevel signal in the data and a limiting effect of annotation noise -particularly around the "To some extent" label.Under lenient evaluation, simple models perform competitively, showing their effectiveness in low-resource settings.