From Topic Networks to Distributed Cognitive Maps: Zipfian Topic Universes in the Area of Volunteered Geographic Information

Table 7

F-scores of classifying ATNs into five classes (CITIES, REGIONS, OTHERS, WP-REGIO-1, and WP-OTHERS-1) by means of SVMs using RBF kernels.

	Measure	all	opt	ext	B1	B2 all	B2 opt	B3 all	B3 opt	B4 all	B4 opt
1	GES	0.598	0.649	0.752	0.200	0.226	0.325	0.182	0.397	0.176	0.294
2	WAL	0.610	0.635	0.707	0.200	0.168	0.222	0.182	0.397	0.158	0.289
3	VEO	0.636	0.706	0.783	0.200	0.213	0.306	0.182	0.397	0.170	0.308
4	wges	0.458	0.576	0.618	0.200	0.311	0.348	0.182	0.397	0.173	0.281
5		0.567	0.673	0.737	0.200	—	—	0.182	0.397	0.173	0.300
6		0.740	0.777	0.854	0.200	0.242	0.440	0.182	0.397	0.181	0.320
7		0.612	0.816	0.875	0.200	—	—	0.182	0.397	0.187	0.340
8		0.559	0.600	0.652	0.200	—	—	0.182	0.397	0.182	0.307
9		0.721	0.811	0.865	0.200	0.240	0.464	0.182	0.397	0.182	0.317
10	NetSimile	0.467	0.507	0.610	0.200	0.494	0.602	0.182	0.397	0.173	0.272
11	ToSi	0.431	0.567	0.585	0.200	-		0.182	0.397	0.179	0.254

Column “all”: F-scores using all features in terms of the respective similarity measure. Column “opt”: using a subset of features detected according to a genetic search. Column “ext”: subset selection according to extended genetic optimization. Additionally, F-scores of random baselines B1, B2, B3, and B4 are displayed, in the latter three cases differentiated for the variants all and opt.