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

Table 12

BiLSTM classification performance using embeddings over the BioText Berkeley dataset.

Embeddings	Accuracy
	,		,		,		,
	Training	Validation	Training	Validation	Training	Validation	Training	Validation
GloVe_C	0.8463	0.8392	0.8310	0.7859	0.8633	0.8551	0.8689	0.8568
GloVe_W	0.8313	0.8122	0.8441	0.7977	0.8653	0.8534	0.8729	0.8152
GloVe_Merged	0.8266	0.8187	0.8171	0.7859	0.8389	0.8240	0.8386	0.8205
CE	0.8438	0.8337	0.8576	0.8018	0.8753	0.8425	0.8766	0.8501
WE	0.8431	0.8335	0.8657	0.8089	0.8616	0.8606	0.8816	0.8459
Merged	0.8505	0.8240	0.8318	0.7969	0.8446	0.8394	0.8468	0.8313

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