An insight analysis and detection of drug-abuse risk behavior on Twitter with self-taught deep learning

Computational Social Networks

Table 4 Drug name and abuse behavior co-occurrence frequency differences between positive tweets and negative tweets

Combo	Pos_count	Neg_count	Ratio_diff (%)	Relative_ratio (%)
Trash high	1131	1387	0.9189	1166.04
Acid trip	547	239	0.4585	1863.66
Acid drop	256	168	0.2127	1603.13
Glass amp	374	3472	0.2060	171.11
Acid take	222	167	0.1838	1509.61
Lean amp	280	2391	0.1610	192.55
Coke high	195	186	0.1602	1343.14
Coke take	185	512	0.1410	646.57
Lean hit	180	745	0.1292	446.33
Acid amp	162	367	0.1262	761.96
Molly pop	160	328	0.1257	823.11
Acid hit	132	138	0.1080	1278.34
Lean pop	121	118	0.0993	1327.49
Acid use	115	238	0.0903	817.21
Shrooms trip	105	55	0.0877	1751.49
Lean high	108	136	0.0876	1147.54
Lean use	159	1479	0.0875	170.62
Blow high	125	675	0.0846	337.93
Upper high	112	382	0.0830	537.16
Dope high	106	509	0.0738	383.46
Coke amp	137	1413	0.0709	146.07
Acid high	65	57	0.0535	1402.44
Coke snort	82	571	0.0513	251.20
Molly amp	88	777	0.0498	183.80
Crack hit	108	1360	0.0479	104.18