Supporting systematic reviews using LDA-based document representations

Table 4 Evaluation on all corpora of SVM classifiers trained with TF-IDF features

	Precision	Recall	F₁-score	Accuracy	ROC	PRC
Youth development
Linear	0.394799	0.686301	0.50125	0.8628422	0.891629	0.508361
RBF	0.0	0.0	0.0	0.89957353	0.13187	0.055498
POLY	0.0	0.0	0.0	0.8995735	0.15324	0.054825
Cigarette packaging
Linear	0.3679999	0.7076923	0.48421052	0.937896	0.939295	0.477252
RBF	0.0	0.0	0.0	0.9588086	0.06347	0.021359
POLY	0.0	0.0	0.0	0.9588086	0.082638	0.021496
Cooking skill
Linear	0.366666	0.482456	0.416666	0.967365	0.922862	0.328018
RBF	0.0	0.0	0.0	0.9758	0.07937	0.012568
POLY	0.0	0.0	0.0	0.97584233	0.51207	0.500
COPD
Linear	0.59523	0.773195	0.67264	0.909	0.927631	0.720464
RBF	0.0	0.0	0.0	0.8792	0.066893	0.064489
POLY	0.0	0.0	0.0	0.8792	0.1139	0.067315
Proton beam
Linear	0.0574	0.07874	0.0664451	0.881734	0.562028	0.063233
RBF	0.0	0.0	0.0	0.9465	0.442163	0.048747
POLY	0.0	0.0	0.0	0.9465	0.482718	0.05424

ISSN: 2046-4053