Multi-class Hierarchical Question Classification for Multiple Choice Science Exams
This work addresses the data bottleneck for question classification in science exams, with incremental improvements in accuracy for educational AI systems.
The authors tackled the problem of limited annotated data for question classification by creating the largest challenge dataset with 7,787 science exam questions and 406 hierarchical domains, and showed that a BERT-based model trained on it achieved a +0.12 MAP gain over previous methods and improved question answering accuracy by +1.7% P@1.
Prior work has demonstrated that question classification (QC), recognizing the problem domain of a question, can help answer it more accurately. However, developing strong QC algorithms has been hindered by the limited size and complexity of annotated data available. To address this, we present the largest challenge dataset for QC, containing 7,787 science exam questions paired with detailed classification labels from a fine-grained hierarchical taxonomy of 406 problem domains. We then show that a BERT-based model trained on this dataset achieves a large (+0.12 MAP) gain compared with previous methods, while also achieving state-of-the-art performance on benchmark open-domain and biomedical QC datasets. Finally, we show that using this model's predictions of question topic significantly improves the accuracy of a question answering system by +1.7% P@1, with substantial future gains possible as QC performance improves.