<![CDATA[Mathematical word problems often contain information that is not entirely relevant to answering the question, causing students to make errors in selecting important information. This study aims to identify relevant quantities in Indonesian mathematical word problem texts using a sequence labeling approach. The dataset consists of 700 problems collected from elementary school mathematics e-books and annotated using the BIO scheme (B-REL, I-REL, and O). The model employed is a combination of BiLSTM and CRF with an 80:20 train–test split. Evaluation is conducted using precision, recall, F1-score, and accuracy, with a focus on the F1-score of the REL class. The results show that the model achieves an F1-score of 0.9726. The dominant error occurs in the misclassification of REL as O, indicating that quantity spans are often truncated across tokens. However, no violations of the BIO scheme are observed, ensuring consistent label sequences. These findings indicate that the combination of BiLSTM and CRF is effective in identifying relevant quantities and has the potential to support the understanding of mathematical word problems.]]>
Copyrights © 2026
