Biomedical Text Classification Using Augmented Word Representation Based on Distributional and Relational Contexts

Table 14

CNN classification performance using embeddings over the PubMed_20k_RCT dataset.

Embeddings	Accuracy
	,		,		,		,
	Training	Validation	Training	Validation	Training	Validation	Training	Validation
GloVe_C	0.7084	0.6841	0.7381	0.7320	0.7305	0.7093	0.7087	0.7197
GloVe_W	0.7110	0.6948	0.7146	0.7148	0.7137	0.7036	0.7147	0.7264
GloVe_Merged	0.6806	0.6737	0.6779	0.6497	0.6828	0.6796	0.6868	0.6844
CE	0.7107	0.6858	0.7451	0.7260	0.7166	0.7128	0.7538	0.7339
WE	0.7166	0.7064	0.7389	0.7263	0.7214	0.7180	0.7464	0.7327
Merged	0.6998	0.6764	0.7289	0.7118	0.6993	0.6888	0.7333	0.7086

Bold means the best performance in the case of each dataset.