Table 9.

Performance comparison against models from related works

Model	F1
Word2vec + CNN [38]	0.708
Entity Attention Bi-LSTM [13]	0.787
Matching the Blanks [36]	0.799
Our Model	0.852

The best scores are in bold