A Comparative Analysis of Fraudulent Recruitment Advertisement Detection Methods in the IoT Environment

Table 5

Experimental results on the unbalanced datasets.

Category	Model	Unbalanced dataset-1				Unbalanced dataset-2
		Traditional learning	RF	0.910	0.919	0.896	0.905	0.928	0.946	0.864	0.896
SVM	0.896		0.898	0.885	0.890	0.924	0.932	0.865	0.892
LR	0.877		0.872	0.874	0.873	0.906	0.874	0.880	0.877
NB	0.830		0.824	0.830	0.826	0.832	0.781	0.826	0.796
Traditional learning + feature extraction	RF+TF-IDF	0.904	0.912	0.890	0.898	0.918	0.940	0.844	0.879
	SVM+TF-IDF	0.911	0.915	0.901	0.906	0.936	0.950	0.882	0.909
	LR+TF-IDF	0.882	0.885	0.870	0.877	0.904	0.912	0.830	0.860
	NB+TF-IDF	0.836	0.834	0.824	0.828	0.861	0.855	0.761	0.791
Deep learning	GRU	0.865	0.861	0.857	0.857	0.912	0.893	0.868	0.879
	Bi-LSTM	0.886	0.881	0.881	0.880	0.915	0.894	0.875	0.883
	TextCNN	0.941	0.942	0.934	0.938	0.960	0.958	0.932	0.944
Deep learning + pretraining	GRU+Word2Vec	0.802	0.796	0.808	0.798	0.855	0.820	0.783	0.800
	GRU+GloVe	0.808	0.804	0.797	0.799	0.859	0.814	0.787	0.799
	Bi-LSTM + Word2Vec	0.889	0.884	0.884	0.884	0.918	0.896	0.885	0.889
	Bi-LSTM + GloVe	0.854	0.848	0.855	0.851	0.901	0.895	0.825	0.853
	TextCNN + Word2Vec	0.938	0.939	0.932	0.936	0.955	0.960	0.920	0.938
	TextCNN + GloVe	0.932	0.937	0.922	0.928	0.949	0.958	0.906	0.928
Category	Model	Unbalanced dataset-3				Unbalanced dataset-4
		Traditional learning	RF	0.944	0.963	0.824	0.875	0.977	0.985	0.768	0.842
SVM	0.940		0.941	0.821	0.867	0.974	0.984	0.733	0.811
LR	0.937		0.879	0.881	0.880	0.969	0.830	0.832	0.831
NB	0.853		0.738	0.840	0.769	0.868	0.613	0.827	0.646
Traditional learning + feature extraction	RF+TF-IDF	0.939	0.962	0.806	0.861	0.976	0.987	0.755	0.831
	SVM+TF-IDF	0.954	0.960	0.861	0.901	0.980	0.984	0.799	0.867
	LR+TF-IDF	0.923	0.920	0.771	0.822	0.965	0.938	0.656	0.724
	NB+TF-IDF	0.890	0.891	0.657	0.705	0.955	0.961	0.541	0.564
Deep learning	GRU	0.947	0.918	0.861	0.885	0.973	0.875	0.833	0.850
	Bi-LSTM	0.946	0.920	0.862	0.888	0.979	0.922	0.837	0.874
	TextCNN	0.969	0.956	0.923	0.938	0.986	0.954	0.896	0.922
Deep learning + pretraining	GRU+Word2Vec	0.916	0.855	0.781	0.809	0.967	0.895	0.733	0.791
	GRU+GloVe	0.923	0.867	0.791	0.791	0.969	0.885	0.775	0.820
	Bi-LSTM+Word2Vec	0.947	0.908	0.869	0.887	0.973	0.905	0.802	0.845
	Bi-LSTM+GloVe	0.932	0.902	0.807	0.845	0.972	0.920	0.758	0.817
	TextCNN+Word2Vec	0.968	0.967	0.900	0.929	0.985	0.981	0.860	0.911
	TextCNN+GloVe	0.949	0.958	0.906	0.928	0.962	0.963	0.879	0.915