Evaluation - task1-intertass-ST4-CR
1233 tweets in qrel, 901 tweets in run.

Accuracy: 0.398

Confusion matrix
Real	Pred.	#
N	N	251
N	NEU	39
N	?	122
N	P	42
N	NONE	37
NEU	N	46
NEU	?	49
NEU	P	38
NEU	NONE	15
NEU	NEU	16
NONE	?	59
NONE	NONE	53
NONE	P	36
NONE	N	50
NONE	NEU	26
P	P	171
P	?	102
P	NEU	24
P	NONE	21
P	N	36

Category evaluation
Cat	P	R	F1
N	0.655	0.511	0.574
NEU	0.152	0.098	0.119
NONE	0.421	0.237	0.303
P	0.596	0.483	0.534

Macro-Precision: 0.456
Macro-Recall: 0.332
Macro-F1: 0.384