Table 3.

Comparing models for in-domain performance

	TP	TN	FP	FN	Ac	Pr	Re	F${}_1$	CK	AUC
LSTM-DA	351	318	60	51	0.858	0.854	0.873	0.863	0.715	0.857
LSTM-W	346	297	103	34	0.824	0.771	0.911	0.835	0.650	0.827
LSTM	309	302	98	71	0.783	0.759	0.813	0.785	0.567	0.784
CNN-W	356	274	126	24	0.808	0.739	0.937	0.826	0.618	0.811
CNN	338	277	123	42	0.788	0.733	0.889	0.804	0.579	0.791
DNN	316	298	102	64	0.787	0.756	0.832	0.792	0.575	0.788
XGB	310	310	90	70	0.795	0.775	0.816	0.795	0.590	0.795
RF	309	316	84	71	0.801	0.786	0.813	0.799	0.603	0.802
SVM-L	281	307	93	99	0.754	0.751	0.739	0.745	0.507	0.753
SVM-N	298	324	76	82	0.797	0.797	0.784	0.790	0.594	0.797
kNN	204	321	79	176	0.673	0.721	0.537	0.615	0.342	0.670
MNB	324	271	129	56	0.763	0.715	0.853	0.778	0.528	0.765
RC	273	322	78	107	0.763	0.778	0.718	0.747	0.524	0.762