Computational text analysis within the Humanities: How to combine working practices from the contributing fields?

Table 1 Experimental results from pilot study on narrative viewpoint classification

		Logistic regression classifier
		Precision	Recall	F-score	Accuracy
(A) Training: Road, Test: Garlan	foc.	0.77	0.71	0.74	0.78
(A) Training: Road, Test: Garlan	non-foc.	0.79	0.84	0.82	0.78
(B) Training: Road, Test: Road	foc.	0.73	0.77	0.75	0.75
(B) Training: Road, Test: Road	non-foc.	0.76	0.73	0.74	0.75
(C) Training: Road + Garlan, Test: Road	foc.	0.76	0.73	0.75	0.79
(C) Training: Road + Garlan, Test: Road	non-foc.	0.81	0.83	0.82	0.79