Evaluating text visualization for authorship analysis

Security Informatics

Table 1 Accuracy and efficiency of task performance

Pt.	Task (related function of the portal)	A(V)	A(~V)	Diff_A	p-value_A	E(V)	E(~V)	Diff_E	p-value_E
1a	Counting category-specific words (feature highlighting)	0.83	0.77	0.06	0.4367	1.45	4.45	−3.00	0.0020 **
1b	Relating category-specific words (feature highlighting)	0.46	0.47	−0.01	0.9286	1.90	2.74	−0.84	0.0084 **
2a	Comparing authors’ use of category-specific features (spider chart)	0.90	0.48	0.42	0.0007 **	1.81	4.61	−2.80	0.0000 **
2b	Comparing authors’ use of sub-category-specific features (spider chart)	0.94	0.68	0.26	0.0090 **	1.55	2.81	−1.26	0.0001 **
2c	Comparing similarity of authors’ writing profiles (spider chart)	1.00	0.81	0.19	0.0118 **	2.00	3.16	−1.16	0.0061 **
3a	Identifying authors’ usage of category-specific features (stylometric heatmap)	0.94	0.32	0.62	0.0000 **	1.52	4.03	−2.51	0.0000 **
3b	Identifying authors’ usage of sub-category-specific features (stylometric heatmap)	0.90	0.19	0.71	0.0000 **	1.32	4.42	−3.1	0.0075 **

Note:
A(V) = Mean accuracy of task performance using visualization.
A(~V) = Mean accuracy of task performance without using visualization.
Diff_A = Mean difference of A(V) – A(~V).
E(V) = Mean efficiency of task performance using visualization (in minutes).
E(~V) = Mean efficiency of task performance without using visualization (in minutes).
Diff_E = Mean difference of E(V) – E(~V) (in minutes).
**p < 0.01.