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

Table 13

CNN-LSTM classification performance using embeddings over the BioText Berkeley dataset.

Embeddings	Accuracy
	,		,		,		,
	Training	Validation	Training	Validation	Training	Validation	Training	Validation
GloVe_C	0.9177	0.8798	0.9099	0.8328	0.9134	0.8768	0.9167	0.8739
GloVe_W	0.9034	0.8856	0.8783	0.8123	0.9021	0.8768	0.9203	0.8358
GloVe_Merged	0.8897	0.8534	0.8620	0.8035	0.8676	0.8358	0.9014	0.8358
CE	0.9202	0.8658	0.8997	0.8501	0.8866	0.8587	0.9192	0.8875
WE	0.9094	0.8218	0.9240	0.8387	0.9173	0.8482	0.9265	0.8718
Merged	0.8984	0.8599	0.8806	0.8603	0.8728	0.8223	0.9175	0.8418

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