[Retracted] Analysis of the Reform Effectiveness of Vocational Skill Identification Based on the New Social Training Model in the New Era under Deep Learning Assessment

Table 3

Skill word extraction performance after adding various types of corpus features with different training set ratios.

Model	Training set proportion
	25%	50%	75%
	Precision, recall, F1, lifting	Precision, recall, F1, lifting	Precision, recall, F1, lifting
Model 0	0.714, 0.7402, 0.7267	0.7549, 0.775, 0.7646	0.7769, 0.7848, 0.7807
Model 1	0.7234, 0.7512, 0.7367, +1.00%	0.7641, 0.7783, 0.7685, +0.42%	0.7841, 0.7882, 0.7859, +0.52%
Model 2	0.7195, 0.75, 0.7341, +0.74%	0.7573, 0.7761, 0.7664, +0.18%	0.7831, 0.7883, 0.7855, +0.48%
Model 3	0.8124, 0.8244, 0.8172, +0.95%	0.8131, 0.8457, 0.8433, +7.87%	0.8511, 0.8567, 0.8532, +7.76%
Model 4	0.8155, 0.8293, 0.8221, +9.54%	0.8451, 0.8539, 0.8489, +8.43%	0.8616, 0.8552, 0.8583, +7.6%
Model 5	0.8157, 0.8223, 0.8191, +1.23%	0.8443, 0.8437, 0.8467, +8.15%	0.8563, 0.8564, 0.8567, +7.6%
Model 6	0.7254, 0.7563, 0.7398, +1.32%	0.7631, 0.7775, 0.7698, +0.44%	0.7818, 0.7891, 0.7852, +0.46%
Model 7	0.8195, 0.8341, 0.8236, +10%	0.8457, 0.8562, 0.8432, +8.45%	0.8634, 0.8617, 0.8623, +8.12%
Model 8	0.8291, 0.8382, 0.8335, +10.7%	0.8491, 0.8534, 0.8517, +8.17%	0.8624, 0.8647, 0.8646, +8.38%