Dataset for Automated Cultural Content Classification in EFL Textbooks: An Imbalance-Aware ELECTRA Framework with LLM-Based Augmentation

This dataset supports the study “Automated Cultural Content Classification in EFL Textbooks: An Imbalance-Aware ELECTRA Framework with LLM-Based Augmentation.” The research dataset consists of 2,195 text segments derived from eight English textbooks used under Indonesia’s Merdeka Curriculum and annotated into four cultural categories: Non-Cultural (NC), Source Culture (SC), International Culture (IC), and Target Culture (TC). Additional files provide data supporting the classification, data partitioning, and augmentation experiments reported in the associated article.