Dual CNN for Relation Extraction with Knowledge-Based Attention and Word Embeddings

Table 2

Precision values for the top 100, 200, and 300 extracted relations.

Accuracy (%)	Top 100	Top 200	Top 300	Average
Word embedding	0.74	0.72	0.64	0.70
Knowledge-based attention model	0.83	0.79	0.72	0.78