Evaluation - task1-intertass-PE
1428 tweets in qrel, 1428 tweets in run.

Accuracy: 0.477

Confusion matrix
Real	Pred.	#
N	NONE	70
N	P	59
N	N	318
N	NEU	25
NEU	N	147
NEU	NONE	92
NEU	P	99
NEU	NEU	29
NONE	NONE	51
NONE	N	49
NONE	P	54
NONE	NEU	5
P	P	283
P	NONE	51
P	N	75
P	NEU	21

Category evaluation
Cat	P	R	F1
N	0.540	0.674	0.599
NEU	0.363	0.079	0.130
NONE	0.193	0.321	0.241
P	0.572	0.658	0.612

Macro-Precision: 0.417
Macro-Recall: 0.433
Macro-F1: 0.425