Research on the method of combining artificial intelligence technology to improve the effectiveness of teaching analytical chemistry in colleges and universities

Figure 1.

Figure 2.

Figure 3.

Figure 4.

Figure 5.

Figure 6.

Figure 7.

Figure 8.

The whole subjects are given the model classification

Master mode	frequency	proportion(%)	Cumulative ratio(%)	Whether it’s an ideal mode
10111	26	8.39%	8.39%	YES
11010	18	5.81%	14.2%	NO
11010	2	0.65%	14.85%	YES
11001	25	8.06%	22.91%	NO
11111	42	13.55%	36.46%	NO
11110	197	63.55%	100%	YES

The ideal master mode is classified

Algorithm	MAP
Count	288
Total number	310
Classification rate	92.2%

The number of models is counted in each class

Attribute master mode	10111	11010	11010	11001	11111	11110	Total
Class A	35	6	0	4	2	2	49
Class B	33	8	0	0	1	4	46
Class C	32	5	0	5	5	5	52
Class D	35	9	0	5	3	5	57
Class E	35	6	3	4	4	5	57
Class F	32	8	0	3	1	5	49

Ablation experiment results

Methods	knowledge	speed	learning	forgetting	Junyi		EdNet-KT1
					AUC	ACC	AUC	ACC
TDINA-K			√	√	0.8068	0.8505	0.7814	0.7247
TDINA-F	√	√	√		0.8672	0.8692	0.7957	0.6942
TDINA-L		√		√	0.8453	0.8484	0.7826	0.724
TDINA-S	√	√	√		0.808	0.8557	0.7915	0.6964
TDINA	√	√	√	√	0.8185	0.8543	0.7769	0.7836

Knowledge state (m represents the overflow value)

Serial number	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18	19	20	21	22	23	24	25	26	27	28	29	30
Epetitions	2	3	3	0	0	0	2	3	3	5	5	6	7	8	8	9	11	12	14	12	13	18	15	15	13	11	10	9	8	7
Interval time	45	12	17	/	/	/	13	8	8	5	5	10	m	51	75	m	36	m	m	37	53	m	7	5	4	5	9	32	8	8

Compare results with the benchmark model in AUC and ACC

Method	Junyi		EdNet-KT1
	AUC	ACC	AUC	ACC
DKT	0.7901	0.8767	0.7517	0.6006
LPKT	0.8073	0.8557	0.7822	0.6709
DKVMN	0.8123	0.8572	0.809	0.7271
HawkesKT	0.804	0.8533	0.7447	0.6359
SAKT	0.7113	0.8659	0.6558	0.6181
SAINT+	0.7388	0.8523	0.6763	0.6507
AKT	0.8091	0.8657	0.7977	0.7122
Ours	0.8216	0.7993	0.8136	0.7207

Data set details

Particulars	Data set
	Junyi	EdNet-KT1
Number of interactions	12120552	33152633
Number of learners	195352	765223
Number of questions	726	14385
Number of knowledge points	42	21
Average number of learner interactions	69.6	45.8
The average number of questions related to knowledge points	19.3	155.24
The average number of knowledge points associated with the test	2	3.6

The results of the trial and the corresponding project response patterns and properties

Subject number	Results	Project response mode	Attribute master mode
1	ABDBCDAADAAAD	1111111111011	11011
2	ABDBCDBACADAB	1111111100101	11011
3	ABBBCDBBDABCB	1111111111100	11110
4	ABDCCDACDABDB	1111110011011	11111
5	ABDBCDBBDACCB	1111101111011	11110
6	ABDBCDBADADBA	1111111011011	11110
......	......	......	......
304	ABDBCDCABCADD	1110101100110	11111
305	ABBBCDCABCABC	0111110100111	11111
306	ABDBCBCBDADAB	1110110111011	11110
307	ABDBCDDDDBDDC	1101110100011	10111
308	ABCBCDABDABDD	1111110110010	11101
309	CBDBCDBDDADDB	0111111010010	11110
310	ADDBCDBBDDBDB	1101110110100	11110

All class properties master probability statistics

mean	A1	A2	A3	A4	A5
Class A	1.0000	0.9148	0.9604	0.9028	0.845
Class B	1.0000	0.9905	0.9756	0.9683	0.8619
Class C	1.0000	0.7717	0.8235	0.8558	0.7408
Class D	1.0000	0.7299	0.7716	0.9505	0.7959
Class E	0.8961	0.8803	0.9278	0.8187	0.7105
Class F	0.9364	0.8824	0.9223	0.8874	0.7055
Totality	0.8864	0.897	0.9998	0.9025	0.7793