Neural Models for Key Phrase Detection and Question Generation
Subramanian, Sandeep, Wang, Tong, Yuan, Xingdi, Zhang, Saizheng, Bengio, Yoshua, Trischler, Adam
–arXiv.org Artificial Intelligence
We propose a two-stage neural model to tackle question generation from documents. First, our model estimates the probability that word sequences in a document are ones that a human would pick when selecting candidate answers by training a neural key-phrase extractor on the answers in a question-answering corpus. Predicted key phrases then act as target answers and condition a sequence-to-sequence question-generation model with a copy mechanism. Empirically, our key-phrase extraction model significantly outperforms an entity-tagging baseline and existing rule-based approaches. We further demonstrate that our question generation system formulates fluent, answerable questions from key phrases. This two-stage system could be used to augment or generate reading comprehension datasets, which may be leveraged to improve machine reading systems or in educational settings.
arXiv.org Artificial Intelligence
May-30-2018
- Genre:
- Research Report (0.82)
- Industry:
- Education
- Assessment & Standards > Student Performance (0.48)
- Curriculum > Subject-Specific Education (0.48)
- Health & Medicine > Therapeutic Area (0.47)
- Education
- Technology: