A Joint Model for Question Answering and Question Generation

Learning to generate natural language workshop, ICML 2017 |

We propose a machine comprehension model that learns jointly to generate questions and answers based on documents. The proposed model uses a sequence-to-sequence framework that encodes the document and generates a question (answer) given an answer (question). Significant improvement in question-answering performance is observed empirically on the SQuAD corpus, confirming our hypothesis that the model benefits from jointly learning to perform both tasks. We believe the joint model offers a new perspective on machine comprehension beyond architectural engineering, and serves as a first step towards autonomous information seeking.