Answering Chinese Elementary School Social Study Multiple Choice Questions
This addresses a domain-specific issue for educational AI applications, but it is incremental as it supplements an existing method with additional modules.
The paper tackles the problem of BERT's poor performance on specific question types like Negation and All-of-the-above in Chinese elementary school social study multiple-choice questions by proposing a framework that cascades BERT with Pre-Processor and Answer-Selector modules, resulting in effective performance improvements.
We present a novel approach to answer the Chinese elementary school Social Study Multiple Choice questions. Although BERT has demonstrated excellent performance on Reading Comprehension tasks, it is found not good at handling some specific types of questions, such as Negation, All-of-the-above, and None-of-the-above. We thus propose a novel framework to cascade BERT with a Pre-Processor and an Answer-Selector modules to tackle the above challenges. Experimental results show the proposed approach effectively improves the performance of BERT, and thus demonstrate the feasibility of supplementing BERT with additional modules.