Evaluation - task1-intertass-ES
1899 tweets in qrel, 1899 tweets in run.

Accuracy: 0.540

Confusion matrix
Real	Pred.	#
N	N	447
N	NONE	84
N	P	205
N	NEU	31
NEU	N	85
NEU	P	93
NEU	NEU	6
NEU	NONE	32
NONE	NONE	80
NONE	P	112
NONE	N	77
NONE	NEU	5
P	P	492
P	N	88
P	NONE	52
P	NEU	10

Category evaluation
Cat	P	R	F1
N	0.641	0.583	0.611
NEU	0.115	0.028	0.045
NONE	0.323	0.292	0.307
P	0.545	0.766	0.637

Macro-Precision: 0.406
Macro-Recall: 0.417
Macro-F1: 0.412