Genre classification is one of the core challenges in Music Information Retrieval. In traditional music, this task becomes particularly difficult due to the need for specialized knowledge and the nuanced stylistic variations across genres. In this work, we focus on Irish traditional music and explore symbolic-domain classification using ABC notation scores. Specifically, we propose a classification system that incorporates two distinct data augmentation strategies: key transposition at the symbolic level and oversampling using SMOTE. Results show that both strategies outperform previous audio-based methods, with symbolic transposition achieving the best overall performance. These findings highlight the effectiveness of symbolic representations and data augmentation in genre classification.