Applying Deep Learning Technologies to Evaluate the Patent Quality with the Collaborative Training

Table 10

Data transfer learning results of 2208 Chinese patent Chinese text data after active learning expansion.

Category	Training set				Test set
	Precision	Recall	-score	Support	Precision	Recall	-score	Support
1	0.95	0.89	0.92	676	0.87	0.79	0.83	35
2	0.68	0.68	0.68	404	0.51	0.54	0.53	18
3	0.65	0.79	0.71	506	0.52	0.64	0.57	21
4	0.55	0.52	0.53	258	0.45	0.42	0.43	10
5	0.45	0.28	0.34	154	0.36	0.22	0.27	8
6	0.43	0.36	0.39	69	0.33	0.67	0.45	3
7	0.56	0.71	0.48	87	0.33	0.33	0.33	3
8	0.70	0.65	0.67	54	1.00	0.50	0.67	2
Microavg	0.73	0.71	0.72	2208	0.62	0.61	0.61	100
Macroavg	0.61	0.61	0.61	2208	0.55	0.51	0.53	100