Table 3.

Classification results obtained over the GXD-caption dataset using different set of features

Classifiers	Precision	Recall	F-measure	Accuracy	Utility-10	Utility-20
NB_AB	0.874 (0.023)	0.656 (0.027)	0.749 (0.022)	0.783 (0.017)	0.866 (0.024)	0.872 (0.023)
RF_AB	0.779 (0.016)	0.768 (0.024)	0.773 (0.017)	0.802 (0.015)	0.765 (0.018)	0.776 (0.017)
NB_CAP	0.831 (0.018)	0.766 (0.024)	0.797 (0.019)	0.808 (0.017)	0.820 (0.019)	0.828 (0.018)
RF_CAP	0.855 (0.019)	0.758 (0.017)	0.804 (0.015)	0.817 (0.014)	0.846 (0.021)	0.853 (0.020)
NB_AB&CAP	0.877 (0.018)	0.816 (0.018)	0.846 (0.014)	0.853 (0.013)	0.870 (0.019)	0.876 (0.018)
RF_AB&CAP	0.876 (0.019)	0.829 (0.015)	0.852 (0.013)	0.858 (0.012)	0.869 (0.020)	0.875 (0.019)

AB indicates using features from titles/abstracts only. CAP indicates using features from captions alone. AB_CAP indicates using features from both captions and titles/abstracts.