Improving Medical Q&A Matching by Augmenting Dual-Channel Attention with Global Similarity

Table 1

Summary of recently proposed text matching models.

Model type	Ref	Datasets	Measures	Best neural network architecture
Representation-based model	9	Insurance QA	Accuracy	HL + CNN
	21	TREC-QA		CNN + GESD
	11	cMedQA	ACC@1	MultiCNNs
	12			Three-level composite CNNs
	13			BiGRU-CNN
Interaction-based model	22	SNLI	Accuracy	ESIM
	16	InsuranceQA	Accuracy	Attentive LSTMs
	17	TREC-QA		ABCNN
	24	WikiQA		IARNN
	25			BiMPM
	18	cMedQA	ACC@1	MCFN