MIT Press

Skip to Main Content

Table 10

Results obtained with different pre-trained word embeddings for the event mention detection only task and for the event detection + classification task.

EMBEDDINGS	MENTION DETECTION ONLY			DETECTION + CLASSIFICATION
EMBEDDINGS	P	R	F1	P	R	F1
Komninos	82.50%	83.53%	82.99%	63.46%	62.93%	63.19%
FastText	79.47%	82.00%	81.25%	62.18%	61.79%	62.02%
GoogleNews	80.60%	81.70%	81.15%	63.20%	62.67%	62.93%
GloVe300	80.07%	79.73%	79.89%	61.87%	59.57%	60.69%
GloVe100	79.30%	80.90%	80.13%	60.23%	58.10%	59.16%
Levy	79.40%	83.13%	81.21%	62.10%	60.83%	61.44%
HistWords Google Aver.	78.93%	77.97%	78.43%	61.03%	57.80%	59.36%
HistWords Google Concat.	79.30%	78.57%	79.03%	62.47%	59.37%	60.87%
HistWords COHA Aver.	77.07%	77.93%	77.48%	56.60%	52.23%	54.34%
HistWords COHA Concat.	79.04%	79.34%	79.20%	59.60%	55.13%	57.29%
HistoLevy	80.47%	83.47%	81.95%	63.20%	61.70%	62.46%
HistoFast	79.90%	81.00%	80.44%	59.80%	55.93%	57.78%
HistoGloVe	80.47%	80.24%	80.51%	62.73%	60.17%	61.42%

EMBEDDINGS	MENTION DETECTION ONLY			DETECTION + CLASSIFICATION
EMBEDDINGS	P	R	F1	P	R	F1
Komninos	82.50%	83.53%	82.99%	63.46%	62.93%	63.19%
FastText	79.47%	82.00%	81.25%	62.18%	61.79%	62.02%
GoogleNews	80.60%	81.70%	81.15%	63.20%	62.67%	62.93%
GloVe300	80.07%	79.73%	79.89%	61.87%	59.57%	60.69%
GloVe100	79.30%	80.90%	80.13%	60.23%	58.10%	59.16%
Levy	79.40%	83.13%	81.21%	62.10%	60.83%	61.44%
HistWords Google Aver.	78.93%	77.97%	78.43%	61.03%	57.80%	59.36%
HistWords Google Concat.	79.30%	78.57%	79.03%	62.47%	59.37%	60.87%
HistWords COHA Aver.	77.07%	77.93%	77.48%	56.60%	52.23%	54.34%
HistWords COHA Concat.	79.04%	79.34%	79.20%	59.60%	55.13%	57.29%
HistoLevy	80.47%	83.47%	81.95%	63.20%	61.70%	62.46%
HistoFast	79.90%	81.00%	80.44%	59.80%	55.93%	57.78%
HistoGloVe	80.47%	80.24%	80.51%	62.73%	60.17%	61.42%