Original GW test run

In questa sezione si ricorda che i parametri della funziona gaussiana (3.21) sono stati settati a:

• centro = $P_i$
• varianza = $P_i \cdot E_i$

Tabella 5.1: Original GW test run - benchmark

	CON Stop List
	ham corpus		spam corpus		prestazioni
(ThreshP, ThreshE)	ham	FN	FN	spam	macro Re	macro Pr
0.02, 0.95	146	4	2	148	0.980	0.980
0.02, 0.90	146	4	2	148	0.980	0.980
0.02, 0.85	122	28	1	149	0.903	0.916
0.02, 0.80	122	28	1	149	0.903	0.916
0.05, 0.80	122	28	2	148	0.900	0.912
0.05, 0.85	122	28	2	148	0.900	0.912
0.05, 0.90	146	4	3	147	0.976	0.976
0.05, 0.95	146	4	3	147	0.976	0.976
0.10, 0.95	146	4	3	147	0.976	0.976
0.10, 0.90	145	5	3	147	0.973	0.973
0.15, 0.90	145	5	3	147	0.973	0.973
0.20, 0.90	143	7	3	147	0.966	0.966
0.30, 0.90	144	6	15	135	0.930	0.931
0.40, 0.90	149	1	111	39	0.626	0.774
	SENZA Stop List
	ham corpus		spam corpus		prestazioni
(ThreshP, ThreshE)	ham	FN	FN	spam	macro Re	macro Pr
0.02, 0.95	147	3	2	148	0.983	0.983
0.02, 0.90	148	2	2	148	0.986	0.986
0.02, 0.85	125	25	1	149	0.913	0.924
0.02, 0.80	124	26	1	149	0.910	0.921
0.05, 0.80	124	26	2	148	0.906	0.917
0.05, 0.85	124	26	2	148	0.906	0.917
0.05, 0.90	125	25	2	148	0.910	0.919
0.05, 0.95	147	3	3	147	0.980	0.980
0.10, 0.95	147	3	3	147	0.980	0.980
0.10, 0.90	148	2	3	147	0.983	0.983
0.15, 0.90	148	2	3	147	0.983	0.983
0.20, 0.90	147	3	3	147	0.980	0.980
0.30, 0.90	144	6	16	394	0.926	0.928
0.40, 0.90	149	1	111	39	0.626	0.774