Security Informatics

Table 16 Top 10 attribute probabilities from each class defined by SMO classifier

From: "Our Little Secret": pinpointing potential predators

*Predatory words*				*Non-predatory words*
*Weight*	*Attribute*	*Weight*	*Attribute*	*Weight*	*Attribute*	*Weight*	*Attribute*
0.89	ok	0.52	were	-0.65	f	-0.45	thought
0.77	call	0.52	there	-0.53	they	-0.44	stuff
0.68	hun	0.48	address	-0.52	kewl	-0.43	long
0.65	leave	0.48	u	-0.52	nice	-0.42	tired
0.65	soon	0.47	mind	-0.50	email	-0.42	my
0.59	sweetie	0.47	older	-0.49	which	-0.42	means
0.57	pm	0.46	very	-0.48	omg	-0.41	idk
0.54	hour	0.43	tight	-0.46	cuz	-0.41	bed
0.52	around	0.43	going	-0.46	a	-0.38	ve
0.52	right	0.42	sure	-0.46	like	-0.38	waiting

Back to article page