A Stacked BiLSTM Neural Network Based on Coattention Mechanism for Question Answering

Table 4

Experimental results of different baselines and our model on the Wiki-QA dataset.

Idx	Model	MAP	MRR
1	LSTM with attention [40]	0.6639	0.6828
2	CNN-Cnt [41]	0.6520	0.6086
3	wGRU-sGRU-Gl2 [42]	0.7537	0.7658
4	wGRU-sGRU-Gl2-Cnt [42]	0.7638	0.7825
5	Stacked BiLSTM	0.7248	0.7333
6	SBiLSTM-coA (cosine + Euclidean)	0.7643	0.7751