A n-Gated Recurrent Unit with review for answer selection

Answer selection is one of the most important techniques in question answering applications since it can improve the user experience to a large extend. To achieve a better answer selection performance, a fundamental approach is to better understand the answers and questions. In this research, motivated by the Ebbinghaus Forgetting Curve which indicated that people should review the knowledge timely to prevent from forgetting, we proposed a new n-Gated Recurrent Unit with Review (nGRUR) model which applies the review mechanism on the gated recurrent unit (GRU). The nGRUR model recurrently reviews the past information after a fixed time distance n, and the process of review is controlled by the new gated units. Experimental results have proven the potential of the proposed model and the quantitative analysis has demonstrated that our model is able to acquire a better sentence level representation.