Question Generation (QG) is the task of generating questions from a given document. Its aims to generate relevant and natural questions, answered by a given answer. However, existing approaches for QG usually fail to utilize the rich text structure that could complement the simple word sequence. Meantime, Cross-entropy based training has notorious limitations, such as exposure bias and inconsistency between train and test measurement. To address the issues, we propose a novel Reinforcement Learning (RL) based Semantics-Reinforced architecture, named SiriQG, for QG task. In SiriQG, we propose a hierarchical attention fusion network, for better modeling of both answer information and passage information by integrating explicit syntactic constraints into attention mechanism, and for better understanding the internal structure of the passage and the connection between answer, which makes it better to fuse different levels of granularity (i.e., passages and questions). Last, we also introduce a hybrid evaluator with using a mixed objective that combines both RL loss and cross-entropy loss to ensure the generation of semantically and syntactically question text. To evaluate the performance, we test our SiriQG model on well-known dataset for QG. Extensive experimental results demonstrated that proposed SiriQG can obtained a significant increase in accuracy comparing existing models based on public dataset, and it consistently outperformed all tested baseline models including the state-of-the-arts (SOTA) techniques.
展开▼
