MCQG: Multitask Approach for Chinese Question Generation and Question Answering
Liang Deng, Jingxin Li, Panhu Qi, Zhenlong Liu, Luchen Zhang · 2022
Question generation and question answering are complicate and require us do more reading comprehension. In this work, we finetuned a multilingual T5(mT5) Transformer in a multitask setting for question generation, question answering using Chinese SQuAD datasets. In the multitask setting, we trained the model to perform answer extraction, question generation and question answering tasks simultaneously. We use one model do all the tasks. To our best knowledge, this is the first academic work that performs end-to-end question generation and question answering from Chinese texts. The experiments evaluations shows that the multitask QG and QA approach is effective and has greater performance than single task setting.