Application of intelligent teaching resource organisation model and construction of performance evaluation model

Qiyi Huang; Yan Zhao

Detalles de publicación y publicación

Applied Mathematics and Nonlinear Sciences

Detalles de la revista

Formato: Revista
eISSN: 2444-8656
Primera edición: 01 Jan 2016
Calendario de la edición: 2 veces al año
Idiomas: Inglés

With the continuous reform and development of teaching mode, intelligent teaching has become one of the important development trends of education in the future. The core idea of intelligent teaching is to use information means to organise corresponding teaching resources, carry out teaching activities, reasonably arrange teaching activities and improve comprehensive benefits [1]. The construction and development of intelligent teaching need dynamic changes are very important to ensure the quality of teaching [2]. Dynamic change also means the unknown, not only the unknown of resources, but also the fundamental change of teaching mode and teaching method. Under the development needs of intelligent teaching, the construction of teaching resources also began to be gradually improved. At present, the organisation mode of intelligent teaching resources is mostly based on traditional teaching resources, which not only has a large construction time span, but also is difficult to meet the needs of dynamic teaching organisation [3]. Many schools have made great efforts in improving the organisation mode of intelligent teaching resources, such as using the advantages of the Internet to improve the number and types of resources, specially designing the dynamic update mechanism of resources to meet the needs of dynamic management, merging, integrating and innovating teaching resources, and finally meeting the needs of users [4]. In performance evaluation, data mining algorithms are also introduced, such as clustering algorithm, establishment of decision tree, etc. [5]. These provide new evaluation strategies for traditional performance evaluation, but although the big data mining algorithm has developed earlier, its application in the field of teaching is still in the initial stage, and the evaluation objects are mostly limited to performance or teachers' teaching, which is relatively single.

Based on this background, this paper studies the organisation mode of intelligent teaching resources and the application performance evaluation model, which is divided into four chapters. The first chapter briefly introduces the development of intelligent teaching and the application background of data mining algorithm. The second chapter mainly introduces the teaching evaluation and big data mining algorithms at home and abroad, and summarises the main application angles and improved algorithms of algorithms at home and abroad. The third chapter constructs the intelligent teaching organisation mode, applies the performance evaluation model, uses the algorithm combining matching ant colony algorithm and association rules to realise data mining of intelligent teaching resources, uses ID3 algorithm to build a decision tree for student evaluation, judges the attributes of evaluation factors, and combines fuzzy mathematics algorithm with analytic hierarchy process to improve the accuracy of performance evaluation. In Chapter 4, the algorithm and model used in this paper are simulated and analysed, and experiments are carried out to determine the effectiveness of the algorithm by analysing the effectiveness and running time of the algorithm. The experimental results show that compared with the traditional algorithm, the maximum frequent set of the algorithm proposed in this paper is significantly lower, the number of mining is higher than the traditional algorithm, and the running time is shortened. It shows high utilisation value in student evaluation and teaching resource performance evaluation.

The innovation of this paper lies in the construction of organisation mode teaching evaluation method. On the one hand, the proposed algorithm can tap the shortcomings of intelligent teaching resources and reflect the subjectivity of students. On the other hand, in the performance evaluation, considering the shortcomings of analytic hierarchy process, combined with the method of fuzzy mathematics, the multi-source method is adopted, and the opinions of different evaluators are comprehensively considered to improve the promotion of performance evaluation on teaching work.

State of the art

In the era of big data, smart teaching has begun to enter the actual teaching activities, and the data of teaching resources is also growing on a large scale. The storage and processing of these resources are very large and unchanged. Teaching evaluation has become more difficult because of the diversity of teaching subjects and the complexity of evaluation objects. Many scholars have made a lot of efforts. For example, Yz et al. [6] proposed a teaching evaluation network with graphical structure, which uses students' scores, comment texts, scores and interpersonal relationships to describe students, courses and other entities. The neural network based on random walk is adopted, and the tensor decomposition based on Bayesian probability tensor decomposition is applied to learn and predict students' scores for non-participating courses. In evaluating English teaching, Chen [7] uses the idea of K-means clustering to analyse the collected original error data, to remove the data considered unreliable by the algorithm, and weighted to average the node measurement data to obtain the final fusion value. By combining big data information fusion with K-means clustering algorithm, the clustering and integration of English teaching ability index parameters are realised, and the corresponding English teaching resource allocation scheme is compiled [8]. Aiming at the shortcomings of bp-8 neural network, a method of on-line optimal quality evaluation based on particle swarm optimisation (AHP) is proposed. In their research, Kong et al. [9] believe that fuzzy comprehensive evaluation is a comprehensive method combining qualitative analysis and quantitative analysis, and designed a Gaussian kernel function optimisation method to accurately determine the weight of teaching quality evaluation indicators. Sun [10] proposed to improve RVM based on feature extraction and empirical mode decomposition of acllmd method, and established classroom theoretical teaching quality evaluation model and experimental teaching quality evaluation model based on RVM algorithm. Huang [11] improved the algorithm based on machine learning technology and Gaussian process, explored the distribution characteristics of samples by using Gaussian mixture function, and improved the classical correlation vector machine model.

To sum up, it can be seen that most of the current research on smart classroom is focussed on specific teaching, covering the importance of smart teaching, teaching strategies, establishment methods, etc., which can rarely be analysed from a technical perspective. In the resource construction, it is mostly described from the aspects of resource diversification and service personalisation, and there is little discussion on the evaluation method. On the other hand, although the algorithms related to data mining continue to emerge, such as classification algorithm, ant algorithm and so on, most of these algorithms belong to theoretical analysis. Their applications in teaching are mostly described from the perspective of database perfection, and they are rarely applied to teaching and teaching evaluation. Therefore, it is of great practical significance to carry out the application of performance evaluation in the organisation mode of intelligent teaching resources.

Methodology

3.1

Mining of intelligent teaching curriculum resources

To effectively promote the development of intelligent teaching, the mining algorithm is applied to curriculum resource mining [12]. The curriculum resources are divided into three aspects: teaching method, content and evaluation. The content input setting item is 3 and the minimum support is 0.1. The data mining algorithm is used to mine the important indicators of teaching resources.

In analysing the association between data, the data mining algorithm based on association rules is generally used, but this kind of algorithm needs to scan the database many times, and the weight may be unreasonable [13]. Therefore, the concept of weighting is considered to be combined with ant colony algorithm. Integrating the matching ant colony algorithm and association rules, the fitness equation of ant colony algorithm is (1) $F = a \cdot | L_{k} | + b \cdot \frac{w sup (L_{k})}{min w sup}$ F = a \cdot \left| {{L_k}} \right| + b \cdot {{w\sup ({L_k})} \over {\min w\sup}} where a represents the data length and b represents the sensitivity coefficient of support. Adjusting ab value can realise the adjustment of weighting degree. The fitness is in positive proportion to the weighted support, and the function is expressed as (2) $w sup (L_{k}) = sup (L_{k}) \prod_{k = 1}^{| L_{k} |} (\forall [i (u) \in L_{k}]) t_{i} [i_{k} (w)]$ w\sup ({L_k}) = \sup ({L_k})\prod\limits_{k = 1}^{\left| {{L_k}} \right|} (\forall [i(u) \in {L_k}]){t_i}[{i_k}(w)]

In order to avoid infrequent itemsets, add constraints, and the equation can be expressed as (3) $w sup (L_{k}) \geq min w sup$ w\sup ({L_k}) \ge \min w\sup

After using Eq. (3), the subsets formed by ant individuals belong to frequent itemsets. In mining association rules, ant individuals add attributes according to the rules. Randomly select the attributes that have not been added, select the attributes according to the state transition rules, and construct different association rules for the obtained attribute items [14]. To ensure the efficiency of path selection and improve the mining effect, explore with prior knowledge and tendency. The rules are as follows: (4) $p_{j}^{t} = {\begin{array}{l} \frac{τ_{j}^{α} η_{j}^{β}}{\sum_{s \in allowed i} τ_{s}^{α} η_{s}^{β}}, j \in allowed t \\ 0, else \end{array}$ p_j^t = \left\{{\matrix{{{{\tau_j^\alpha \eta_j^\beta} \over {\sum\limits_{s \in allowedi} \tau_s^\alpha \eta_s^\beta}},j \in allowedt} \hfill \cr {0,else} \hfill \cr}} \right. where j represents the attribute items not accessed by ants, and η represents heuristic information. In association rules, it is necessary to analyse which attributes meet the requirement set, and the pheromone corresponds to the attribute item, so the pheromone needs to be updated [15]. The attribute item pheromone on the rule is updated by the global update method, and the is expressed as (5) $τ_{i} = (1 - p) t_{i} + ρ Δ t_{i}$ {\tau_i} = (1 - p){t_i} + \rho \Delta {t_i} where Δt represents the pheromone increase and i represents the element. The information update of each ant adopts the local update rule, and the equation is: (6) $τ_{i} = (1 - ξ) τ_{i} + ξ τ_{0}$ {\tau_i} = (1 - \xi){\tau_i} + \xi {\tau_0} where ξ represents the proportion of pheromone disappearance, ranging from 0 to 1. The matrix vector is used to connect the resource storage results and reduce the number of scans. It is assumed that each weight of the database item set is expressed as (7) $W (X) = \prod_{l}^{| X |} W (i_{j}) / \sum_{l}^{| X |} W (i_{j})$ W(X) = \prod\limits_l^{\left| X \right|} W({i_j})/\sum\limits_l^{\left| X \right|} W({i_j}) where W (X) represents the weight, i_j ∈ X. The frequency of occurrence of itemset X is expressed as (8) $Fre (X) = | T : X \subseteq T, T \subseteq DB | / | DB |$ Fre(X) = \left| {T:X \subseteq T,T \subseteq DB\left| / \right|DB} \right| where Fre(X) is the frequency. The data item in the error belongs to the item set and is represented by 1, otherwise it is 0. To minimise the number of candidate sets, those with vector degree less than K are removed.

3.2

Data mining of students' teaching evaluation

In the performance evaluation model, students are the main body that cannot be ignored. In the evaluation model, ID3 algorithm is proposed to build a decision tree to judge the attributes of evaluation factors [16]. In the evaluation of the organisation mode of intelligent teaching resources, the teaching situation is distinguished by classification. Each classification will produce some rules and find the factors affecting teaching [17].

The channel model is established. The symbol sent by the signal is represented by u_i, the symbol received is represented by v_j and the conditional probability P represents the transmission probability. The related equation is: (9) $\sum P < v_{j} | u_{i} > = 1, i = 1, 2, \dots, r$ \sum P < {v_j}\left| {{u_i}} \right. > = 1,i = 1,2, \ldots,r

Given the triple channel, the rate channel model is determined. The occurrence probability of the message constitutes the sample space, which is expressed as (10) $[\begin{matrix} U \\ P \end{matrix}] = [\begin{matrix} u_{1}, u_{2}, \dots, u_{r} \\ P (u_{1}), P (u_{2}), \dots, P (u_{r}) \end{matrix}]$ \left[ {\matrix{{\rm{U}} \cr {\rm{P}} \cr}} \right] = \left[ {\matrix{{{u_1},{u_2}, \ldots,{u_r}} \cr {P({u_1}),P({u_2}), \ldots,P({u_r})} \cr}} \right] where U represents the input signal letter and P represents the output signal. The number of messages contained after the occurrence of the message reflects the uncertainty, and the equation is: (11) $log \frac{1}{ρ (u_{i})} = - log p (u_{i})$ \log {1 \over {\rho ({u_i})}} = - \log p({u_i}) where, the unit of information is bit. After the source sends a message, the amount of information reflects the average certainty, and the equation is: (12) $Hu = \sum_{i} p (u_{i}) {log}_{2} \frac{1}{p (u_{i})}$ Hu = \sum\limits_i p({u_i})\mathop {\log}\nolimits_2 {1 \over {p({u_i})}} where H(u) represents information entropy. When H(u) is 0, there is only one possibility. If the probability of occurrence is the same, H(u) reaches the maximum value.

There is interference signal in general channel, so it is necessary to measure this uncertain signal [18]. The average uncertainty formula is calculated as (13) $H < U | V_{j} > = \sum_{i} p < u_{i} | v_{j} > {log}_{2} \frac{1}{p (u_{i}) | v_{j}}$ H < U\left| {{V_j}} \right. > = \sum\limits_i p < {u_i}\left| {{v_j}} \right. > \mathop {\log}\nolimits_2 {1 \over {p({u_i})\left| {{v_j}} \right.}}

Now, calculate the expected value in the symbol set V_j to obtain the conditional entropy. The equation for that is (14) $H < U | V > = \sum_{j} p (v_{j}) \sum_{i} p < u_{i} | v_{j} > {log}_{2} \frac{1}{p (u_{i}) | v_{j}}$ H < U\left| V \right. > = \sum\limits_j p({v_j})\sum\limits_i p < {u_i}\left| {{v_j}} \right. > \mathop {\log}\nolimits_2 {1 \over {p({u_i})\left| {{v_j}} \right.}} where the conditional entropy is also the degree of doubt. If one-to-one corresponds to the channel, this value is 0. For a given channel, mutual information belongs to n type function. There must be a probability to maximise mutual information. This maximum value is the channel capacity. The formula is: (15) $C = Max {I (U, V)}$ C = Max\{I(U,V)\} where C represents the channel capacity. Assuming that the data set is represented by U, the information entropy can be expressed as (16) $I (U) = \sum P (U_{l}) {log}_{2} P (U_{l})$ {\rm{I}}(U) = \sum P({U_l}){\log_2}P({U_l})

Conditional entropy is expressed as (17) $E (A) = \sum_{j} P (V_{1}) \sum P (U_{l} | V_{j}) {log}_{2} \frac{1}{P (U_{l} | V_{j})}$ E(A) = \sum\limits_j P({V_1})\sum P({U_l}|{V_j})\mathop {\log}\nolimits_2 {1 \over {P({U_l}|{V_j})}}

3.3

Performance evaluation of teaching resource organisation model

In the performance evaluation of teaching resource organisation mode, the analytic hierarchy process is used to realise and select the evaluation index and to construct the evaluation system [19]. According to a certain logical relationship, the more complex evaluation system compares the indicators, constructs the consistency judgement matrix, obtains the feature vector, and obtains the weight of each index after normalisation [20]. Practically, due to the influence of personal preference, the consistency test of judgement matrix cannot be carried out. Therefore, fuzzy mathematics is proposed to realise comprehensive evaluation.

The fuzzy mathematical evaluation model covers factor set, decision set and single factor decision-making. When the single factor decision-making is fuzzy mapping, there are: (18) $f : U \to F (V), u_{i} \to f (u_{i}) \equiv (r_{i 1}, r_{i 2}, \dots r_{im}) \in F (V)$ f:U \to F(V),{u_i} \to f({u_i}) \equiv ({r_{i1}},{r_{i2}}, \ldots {r_{im}}) \in F(V) r belongs to fuzzy matrix, and a fuzzy transformation can be obtained according to the following equation: (19) $\begin{array}{l} T_{R} : F (U) & \to F (V) \\ A & \to T_{R} (\tilde{A}) = A \cdot R \end{array}$ \matrix{{{T_R}:F(U)} \hfill & {\to F(V)} \hfill \cr A \hfill & {\to {T_R}(\tilde A) = A \cdot R} \hfill \cr}

Input weight classification and output comprehensive decision can be expressed as an equation, which is (20) $b_{j} = \lor_{i = 1}^{n} (w_{i} \land r_{ij})$ {b_j} = \mathop \vee \limits_{i = 1}^n ({w_i} \wedge {r_{ij}}) where w_i represents the weight distribution index and j is a natural number. In the performance evaluation, according to the group evaluation results, calculate the weight of secondary indicators to form a secondary indicator fuzzy matrix, which can be expressed as (21) $R_{i} = [\begin{matrix} r_{t 11}, r_{t 12}, \dots, r_{t 1 n} \\ r_{t 21}, r_{t 22}, \dots, r_{t 2 n} \\ \dots \dots \\ r_{tj 1}, r_{tj 2}, \dots, r_{tjn} \end{matrix}]$ {R_i} = \left[ {\matrix{{{r_{t11}},{r_{t12}}, \ldots,{r_{t1n}}} \cr {{r_{t21}},{r_{t22}}, \ldots,{r_{t2n}}} \cr {\ldots \ldots} \cr {{r_{tj1}},{r_{tj2}}, \ldots,{r_{tjn}}} \cr}} \right] where i represents the number of indicators, t represents the secondary indicators and n represents the grade. Considering that the weight distribution is affected by many factors, the primary index matrix can be obtained according to the multiplication rules, which can be expressed as (22) $E = [B_{n}] = [W_{n} \cdot R_{n}]$ E = [{B_n}] = [{W_n} \cdot {R_n}] where E represents the primary index and represents the fuzzy evaluation matrix. According to the weight set and evaluation matrix, the comprehensive evaluation is obtained by using the maximum and minimum principle. The formula is: (23) $X = W \cdot E$ X = W \cdot E where X represents the comprehensive evaluation matrix. The matrix is normalised. Due to individual differences, the importance indicators of indicators are not exactly the same, so consistency test is needed. Suppose there are m experts and n evaluation indicators. The normalised matrix is expressed as (24) $B_{k} = (b_{kj}^{k})_{mn}$ {B_k} = (b_{kj}^k{)_{mn}} where b represents the vector, and the average vector can be expressed as (25) $b_{k} = \frac{1}{n} \sum_{j}^{n} b_{j}^{(k)}$ {b_k} = {1 \over n}\sum\limits_j^n b_j^{(k)} where k is a natural number. The fuzzy relation matrix is obtained according to the consistency test. The matrix has no transitivity and needs to be operated for many times [21]. Without detailed analysis of expert opinions, complex calculations are not required [22]. Assuming that the judgement matrix is represented by A and the ranking vector is represented by W, the statistical deviation of the judgement matrix can be expressed as (26) $u = \frac{1}{n (n - 1)} \sum_{1 \leq i \leq j \leq n} δ_{ij}^{2}$ u = {1 \over {n(n - 1)}}\sum\limits_{1 \le i \le j \le n} \delta_{ij}^2 where u represents the consistency inspection measure. The smaller the value, the better the consistency inspection. To ensure the accuracy of the results, the grades are expressed by numerical values and divided into four grades. The comprehensive evaluation results can be expressed as follows (27) $Q = X \cdot P^{T}$ Q = X \cdot {P^T}

The maximum and minimum method of fuzzy mathematics is used to calculate and normalise [23], and the equation is: (28) $V_{i} (u_{0}) = \underset{lin}{\lor} (V_{j} u_{0})$ {V_i}({u_0}) = \mathop \vee \limits_{lin} ({V_j}{u_0}) where u means relative subordinate to V. When evaluating multiple indicators, the single calculation score is unreasonable and needs fuzzy processing. Assuming that there are n indicators and m secondary indicators for evaluation, the larger the performance score, the better the indicators. The equation is expressed as (29) $r_{ij} = {\begin{matrix} 1, S_{ij} 〉 S_{ip} \\ \frac{S_{ij} - S_{if}}{S_{ip} - S_{if}}, S_{if} 〈 S_{ij} 〈 S_{ip} \\ 0, S_{if} 〉 S_{ij} \end{matrix}$ {r_{ij}} = \left\{{\matrix{{1,{S_{ij}}\rangle {S_{ip}}} \hfill \cr {{{{S_{ij}} - {S_{if}}} \over {{S_{ip}} - {S_{if}}}},{S_{if}}\langle {S_{ij}}\langle {S_{ip}}} \hfill \cr {0,{S_{if}}\rangle {S_{ij}}} \hfill \cr}} \right. where r_ij represents the membership degree of the index and s represents the limit of the index value. Establish a standard scheme, and the equation of the maximum membership degree of excellent performance is (30) $\bar{G} = {(g_{1}, g_{2}, \dots, g_{n})}^{T}$ \overline G = {\left({{g_1},{g_2}, \ldots,{g_n}} \right)^T}

There are various factors affecting performance evaluation, so it is not possible to make an overall analysis by virtue of good or bad. Moreover, many indicators are highly subjective and difficult to be described directly by data [24]. In analysing the application of the organisation mode of intelligent teaching resources, the investigation indicators are divided into four categories: resource scale, network performance, experience and rich content. The scale of resources reflects the richness of resources. At present, smart classroom teaching resources directly affect students, and efficiency is a factor that must be considered. In this evaluation, the selected resource scale indicators cover the total pages and links, and web analyst is used to capture data directly. The network performance reflects the technical complaints, and the evaluation indicators include integrity, link indicators and return. The experience reflects the subjective feeling of students' application, including download speed, friendly interface and connection speed. The rich content reflects the content of resources, including public content, shared content and downloadable content. The weight of each index is obtained according to the analytic hierarchy process. It is sorted by calculating the importance distance. In the construction process, consistency test is required.

Result analysis and discussion

4.1

Simulation test

Comparing the resource mining algorithm proposed in this paper with the classical Apriori algorithm, the minimum weighted support is (0.1, 0.2), the confidence is (0.2, 0.3), the running time of this algorithm is 0.625, the number of mining is 1243, the running time of classical algorithm is 17.53 and the number of mining is 1063. It shows that the running time of the algorithm is significantly shortened and the number of resource mining is also significantly improved. The stability analysis results of the algorithm are shown in Table 1. It can be seen from the data in the table that under the same minimum support conditions, compared with the basic Apriori algorithm, the maximum frequent set of the algorithm proposed in this paper is significantly lower. With the increase of minimum support, this advantage is still obvious, indicating that the stability of the algorithm in this paper is strong.

Table 1

Comparison of algorithm stability

Minimum support		0.05	0.10	0.20	0.30	0.40	0.50

Number of maximum frequent set	Apriori algorithm	6.65	4.45	3.82	3.26	2.36	1.47
Number of maximum frequent set	Algorithm of this paper	4.03	3.17	2.00	1.42	1.24	0.75
Execution unit time (ms)	Apriori algorithm	5.22	4.92	3.54	2.51	1.72	1.64
Execution unit time (ms)	Algorithm of this paper	3.00	2.04	1.85	1.41	0.85	0.72

In the teaching evaluation, the teaching evaluation data of one academic year is selected as the data source, and the speciality with strong speciality is selected. The number of included samples is 50. The evaluation of teachers belongs to multi-level analysis. The evaluation indicators should select the data with important attributes. The evaluation attributes include gender, professional title, attitude, method, effect and content, with a total of six attributes. The data to be analysed is taken as the original data and sorted into the same database as the mining data source. There may be gaps in these data, so these inaccurate or correct data are proposed. In the evaluation, all numbers are required to be of continuous values, so it is necessary to classify them. According to the actual situation, the gender is divided into two grades, the professional title is divided into four grades, the attitude is divided into four grades, and the method, content and effect are divided into four grades. The calculated professional title information gain is 0.321, gender information gain is 0.103, attitude information gain is 0.576, content information gain is 0.532, method information gain is 0.514 and effect information gain is 0.428 The most far-reaching factor affecting teaching is teaching attitude, which has no obvious relationship with gender and professional title. Teaching attitude will directly affect the quality of teaching. Teachers' professional titles basically reflect their professional level. In this calculation data, it can be seen that the impact of professional titles on teaching is not very high. However, based on teaching attitude, teachers with high professional titles score better. It can be seen that through the data mining algorithm, we can deeply mine the results of students' teaching evaluation, find a variety of factors affecting the teaching quality and the correlation between them, which plays a certain role in improving the teaching quality.

4.2

Application performance evaluation analysis of intelligent teaching resource organisation model

Decompose the data according to the analytic hierarchy process and divide the data into different levels. Four primary indicators are used as the quasi side layers, which are resource scale, network performance, experience and rich content. There are 11 secondary indicators as the decision-making level, which are the total page of resources, link status, resource integrity, link indicators, web page return status, download speed, friendly interface, connection speed, public content, shared content and downloadable content. According to the hierarchical relationship, the upper level is the goal, and the relative importance of the lower level is the weight. Combined with the above algorithm, the indexes that are difficult to quantify are introduced into the matrix. Now select SPSS software to continue the analysis and standardise the data. All data range is from 1 to 100. Then convert the hundred point system and weigh it with the weight of secondary indicators to calculate the weight of primary indicators. Table 2 shows the score of primary indicators of performance evaluation.

Table 2

Scores of primary indicators of performance evaluation

Evaluating indicator	Resource scale	Network performance	Experience situation	Rich content

A	86.4	69.4	77.4	79.3
B	88.2	71.3	82.6	81.4
C	81.6	69.2	79.5	77.3
D	89.4	66.7	81.5	78.5

The membership matrix is established according to the weight, the closeness between excellent performance and ideal performance is obtained, and then the ranking result is obtained. It can be seen from the data in Table 2 that among the primary indicators, the resource scale has the highest score and the network performance has the lowest score. These data show that the current smart teaching website scale construction basically meets the needs of students. Although the network performance develops rapidly, it also has its own shortcomings. The content and resource experience are good, and the resource recommendation score is low, which needs to be improved. The weight of website content is not the largest, indicating that the content needs to be further enriched, which is also the material basis to meet the needs of intelligent teaching.

Conclusion

In the context of big data, smart teaching has become one of the inevitable development directions of teaching reforms. At present, many colleges and universities have reformed based on previous information-based teaching, and teaching resources are constantly enriched. However, due to algorithm reasons, most of the previous teaching performance evaluation started from a single point of view, and most of them were teachers' evaluation of students. Based on this, this paper studies the application performance evaluation model of intelligent teaching resource organisation model. Based on using big data mining algorithm to mine resource information, taking students as the main body and using decision tree model to establish student evaluation method, the classical analytic hierarchy process is selected for performance evaluation. Considering that some indicators are difficult to describe accurately, the comprehensive evaluation is carried out in combination with the concept of fuzzy mathematics. The experimental analysis proves the effectiveness and stability of the proposed algorithm, and it can shorten the running time. Through the weight calculation, the main factors affecting the performance can be obtained, and the deficiencies in teaching can be found. It should be pointed out that smart teaching is still in the development stage, and will continue to develop in the future. There will be a series of changes and new contents in resource construction. With the continuous progress and maturity of big technology, more technologies will be applied to the performance evaluation of intelligent teaching. Moreover, the data mined in this paper is not many, and the more in-depth data related to teaching needs further research.

Figuras y tablas

Scores of primary indicators of performance evaluation

Evaluating indicator	Resource scale	Network performance	Experience situation	Rich content

A	86.4	69.4	77.4	79.3
B	88.2	71.3	82.6	81.4
C	81.6	69.2	79.5	77.3
D	89.4	66.7	81.5	78.5

Comparison of algorithm stability

Minimum support		0.05	0.10	0.20	0.30	0.40	0.50

Number of maximum frequent set	Apriori algorithm	6.65	4.45	3.82	3.26	2.36	1.47
Number of maximum frequent set	Algorithm of this paper	4.03	3.17	2.00	1.42	1.24	0.75
Execution unit time (ms)	Apriori algorithm	5.22	4.92	3.54	2.51	1.72	1.64
Execution unit time (ms)	Algorithm of this paper	3.00	2.04	1.85	1.41	0.85	0.72

Referencias

[1] Tingle E, Groen J. Book Review: Seeking Wisdom in Adult Teaching and Learning: An Autoethnographic Inquiry, by Fraser, W[J]. Adult Education Quarterly, 2019, 70(3):5–8. TingleE GroenJ Book Review: Seeking Wisdom in Adult Teaching and Learning: An Autoethnographic Inquiry, by Fraser, W[J] Adult Education Quarterly 2019 70 3 5 8 10.1177/0741713619889579 Search in Google Scholar

[2] Paul M, Jena L K, Sahoo K. Workplace Spirituality and Workforce Agility: A Psychological Exploration Among Teaching Professionals[J]. Journal of Religion and Health, 2020, 59(1):135–153. PaulM JenaL K SahooK Workplace Spirituality and Workforce Agility: A Psychological Exploration Among Teaching Professionals[J] Journal of Religion and Health 2020 59 1 135 153 10.1007/s10943-019-00918-331549287 Search in Google Scholar

[3] Villanueva KA, Brown SA, Pitterson NP, et al. Teaching evaluation practices in engineering programs: Current approaches and usefulness[J]. International Journal of Engineering Education, 2017, 33(4):1317–1334. VillanuevaKA BrownSA PittersonNP Teaching evaluation practices in engineering programs: Current approaches and usefulness[J] International Journal of Engineering Education 2017 33 4 1317 1334 Search in Google Scholar

[4] Xia X, Yan J. Construction of Music Teaching Evaluation Model Based on Weighted Nave Bayes[J]. Scientific Programming, 2021, 2021:1–9. XiaX YanJ Construction of Music Teaching Evaluation Model Based on Weighted Nave Bayes[J] Scientific Programming 2021 2021 1 9 10.1155/2021/7196197 Search in Google Scholar

[5] Ghosh B. Spatial mapping of groundwater potential using data driven evidential belief function, knowledge based analytic hierarchy process and an ensemble approach[J]. Environmental Earth Sciences, 2021, 80(18):625. GhoshB Spatial mapping of groundwater potential using data driven evidential belief function, knowledge based analytic hierarchy process and an ensemble approach[J] Environmental Earth Sciences 2021 80 18 625 10.1007/s12665-021-09921-y Search in Google Scholar

[6] Yz A, Hao LA, Ping QA, et al. Heterogeneous teaching evaluation network based offline course recommendation with graph learning and tensor factorization - Science Direct[J]. Neurocomputing, 2020, 415:84–95. YzA HaoLA PingQA Heterogeneous teaching evaluation network based offline course recommendation with graph learning and tensor factorization - Science Direct[J] Neurocomputing 2020 415 84 95 10.1016/j.neucom.2020.07.064 Search in Google Scholar

[7] Chen Z. Using Big Data Fuzzy K-Means Clustering and Information Fusion Algorithm in English Teaching Ability Evaluation[J]. Complexity, 2021, 2021(5):1–9. ChenZ Using Big Data Fuzzy K-Means Clustering and Information Fusion Algorithm in English Teaching Ability Evaluation[J] Complexity 2021 2021 5 1 9 10.1155/2021/2574025 Search in Google Scholar

[8] Jiang L, Wang X. Optimization of Online Teaching Quality Evaluation Model Based on Hierarchical PSO-BP Neural Network[J]. Complexity, 2020, 2020(7):1–12. JiangL WangX Optimization of Online Teaching Quality Evaluation Model Based on Hierarchical PSO-BP Neural Network[J] Complexity 2020 2020 7 1 12 10.1155/2020/6647683 Search in Google Scholar

[9] Kong H, Fan H, Yan Z, et al. Design of teaching quality evaluation model based on fuzzy mathematics and SVM algorithm[J]. Journal of Intelligent and Fuzzy Systems, 2018, 35:1–9. KongH FanH YanZ Design of teaching quality evaluation model based on fuzzy mathematics and SVM algorithm[J] Journal of Intelligent and Fuzzy Systems 2018 35 1 9 10.3233/JIFS-169663 Search in Google Scholar

[10] Sun Q. Evaluation model of classroom teaching quality based on improved RVM algorithm and knowledge recommendation[J]. Journal of Intelligent and Fuzzy Systems, 2021, 40(2):2457–2467. SunQ Evaluation model of classroom teaching quality based on improved RVM algorithm and knowledge recommendation[J] Journal of Intelligent and Fuzzy Systems 2021 40 2 2457 2467 10.3233/JIFS-189240 Search in Google Scholar

[11] Huang W. Simulation of English teaching quality evaluation model based on gaussian process machine learning[J]. Journal of Intelligent and Fuzzy Systems, 2021, 40(2):2373–2383. HuangW Simulation of English teaching quality evaluation model based on gaussian process machine learning[J] Journal of Intelligent and Fuzzy Systems 2021 40 2 2373 2383 10.3233/JIFS-189233 Search in Google Scholar

[12] Long Y. Research on art innovation teaching platform based on data mining algorithm[J]. Cluster Computing, 2019, 22(2):1–7. LongY Research on art innovation teaching platform based on data mining algorithm[J] Cluster Computing 2019 22 2 1 7 10.1007/s10586-018-2461-z Search in Google Scholar

[13] Tong-Zhigang. Research on decision support system of sports assistant teaching and training based on association rules and support vector machine[J]. Journal of Intelligent and Fuzzy Systems, 2021(2):1–12. Tong-Zhigang Research on decision support system of sports assistant teaching and training based on association rules and support vector machine[J] Journal of Intelligent and Fuzzy Systems 2021 2 1 12 10.3233/JIFS-219035 Search in Google Scholar

[14] Mokhtari NA, Ghezavati V. Integration of efficient multi-objective ant-colony and a heuristic method to solve a novel multi-objective mixed load school bus routing model[J]. Applied Soft Computing, 2018, 68:92–109. MokhtariNA GhezavatiV Integration of efficient multi-objective ant-colony and a heuristic method to solve a novel multi-objective mixed load school bus routing model[J] Applied Soft Computing 2018 68 92 109 10.1016/j.asoc.2018.03.049 Search in Google Scholar

[15] Goel R, Maini R. A hybrid of Ant Colony and firefly algorithms (HAFA) for solving vehicle routing problems[J]. Journal of Computational Science, 2018, 25(5):28–37. GoelR MainiR A hybrid of Ant Colony and firefly algorithms (HAFA) for solving vehicle routing problems[J] Journal of Computational Science 2018 25 5 28 37 10.1016/j.jocs.2017.12.012 Search in Google Scholar

[16] Bhuvaneswari G, Manikandan G. A novel machine learning framework for diagnosing the type 2 diabetics using temporal fuzzy ant miner decision tree classifier with temporal weighted genetic algorithm[J]. Computing, 2018, 100(8):1–14. BhuvaneswariG ManikandanG A novel machine learning framework for diagnosing the type 2 diabetics using temporal fuzzy ant miner decision tree classifier with temporal weighted genetic algorithm[J] Computing 2018 100 8 1 14 10.1007/s00607-018-0599-4 Search in Google Scholar

[17] Gu Z, He C. Application of Fuzzy Decision Tree Algorithm Based on Mobile Computing in Sports Fitness Member Management[J]. Wireless Communications and Mobile Computing, 2021, 2021(6):1–10. GuZ HeC Application of Fuzzy Decision Tree Algorithm Based on Mobile Computing in Sports Fitness Member Management[J] Wireless Communications and Mobile Computing 2021 2021 6 1 10 10.1155/2021/4632722 Search in Google Scholar

[18] Li L, Dai S, Cao Z, et al. Using improved gradient-boosted decision tree algorithm based on Kalman filter (GBDT-KF) in time series prediction[J]. The Journal of Supercomputing, 2020, 76(9):6887–6900. LiL DaiS CaoZ Using improved gradient-boosted decision tree algorithm based on Kalman filter (GBDT-KF) in time series prediction[J] The Journal of Supercomputing 2020 76 9 6887 6900 10.1007/s11227-019-03130-y Search in Google Scholar

[19] Kamaruzzaman SN, Lou E, Wong PF, et al. Developing weighting system for refurbishment building assessment scheme in Malaysia through analytic hierarchy process (AHP) approach[J]. Energy Policy, 2018, 112:280–290. KamaruzzamanSN LouE WongPF Developing weighting system for refurbishment building assessment scheme in Malaysia through analytic hierarchy process (AHP) approach[J] Energy Policy 2018 112 280 290 10.1016/j.enpol.2017.10.023 Search in Google Scholar

[20] Chen Y, Bouferguene A, Al-Hussein M. Analytic Hierarchy Process–Simulation Framework for Lighting Maintenance Decision-Making Based on the Clustered Network[J]. Journal of Performance of Constructed Facilities, 2018, 32(1):04017114. ChenY BoufergueneA Al-HusseinM Analytic Hierarchy Process–Simulation Framework for Lighting Maintenance Decision-Making Based on the Clustered Network[J] Journal of Performance of Constructed Facilities 2018 32 1 04017114 10.1061/(ASCE)CF.1943-5509.0001101 Search in Google Scholar

[21] Ooi J, Promentilla M, Tan RR, et al. Integration of Fuzzy Analytic Hierarchy Process into multi-objective Computer Aided Molecular Design[J]. Computers & Chemical Engineering, 2018, 109(JAN.4):191–202. OoiJ PromentillaM TanRR Integration of Fuzzy Analytic Hierarchy Process into multi-objective Computer Aided Molecular Design[J] Computers & Chemical Engineering 2018 109 JAN. 4 191 202 10.1016/j.compchemeng.2017.11.015 Search in Google Scholar

[22] Evaluation of Aqueous Phase Trapping in Shale Gas Reservoirs Based on Analytic Hierarchy Process[J]. Energy And Fuels, 2021, 35(2):1389–1397. Evaluation of Aqueous Phase Trapping in Shale Gas Reservoirs Based on Analytic Hierarchy Process[J] Energy And Fuels 2021 35 2 1389 1397 10.1021/acs.energyfuels.0c03867 Search in Google Scholar

[23] Zuhri M, Shaharuzaman MA, Mansor MR, et al. The weighting of product design specification for a composite side-door impact beam using the analytic hierarchy process method[J]. International Journal of Materials and Product Technology, 2019, 59(1):63–72. ZuhriM ShaharuzamanMA MansorMR The weighting of product design specification for a composite side-door impact beam using the analytic hierarchy process method[J] International Journal of Materials and Product Technology 2019 59 1 63 72 10.1504/IJMPT.2019.10021829 Search in Google Scholar

[24] Ener E, Ener E, Davraz A. Groundwater potential mapping by combining fuzzy-analytic hierarchy process and GIS in Beyehir Lake Basin, Turkey[J]. Arabian Journal of Geosciences, 2018, 11(8):187. EnerE EnerE DavrazA Groundwater potential mapping by combining fuzzy-analytic hierarchy process and GIS in Beyehir Lake Basin, Turkey[J] Arabian Journal of Geosciences 2018 11 8 187 10.1007/s12517-018-3510-x Search in Google Scholar

Applied Mathematics and Nonlinear Sciences

Application of intelligent teaching resource organisation model and construction of performance evaluation model

Qiyi Huang y
Qiyi Huang y
Yan Zhao
Yan Zhao

Publicado en línea: 24 Aug 2022

Volumen & Edición: AHEAD OF PRINT

Páginas: -

Recibido: 31 Mar 2022

Aceptado: 30 May 2022

Detalles de publicación y publicación

Applied Mathematics and Nonlinear Sciences

Vista previa del PDF

Artículo

Figuras y tablas

Scores of primary indicators of performance evaluation

Comparison of algorithm stability

Referencias

Artículos Recientes

Planifique su conferencia remota con Sciendo

Applied Mathematics and Nonlinear Sciences

Application of intelligent teaching resource organisation model and construction of performance evaluation model

Qiyi Huang yQiyi Huang yYan ZhaoYan Zhao

Publicado en línea: 24 Aug 2022

Volumen & Edición: AHEAD OF PRINT

Páginas: -

Recibido: 31 Mar 2022

Aceptado: 30 May 2022

Detalles de publicación y publicación

Applied Mathematics and Nonlinear Sciences

Vista previa del PDF

Artículo

Figuras y tablas

Scores of primary indicators of performance evaluation

Comparison of algorithm stability

Referencias

Artículos Recientes

Planifique su conferencia remota con Sciendo

Qiyi Huang y
Qiyi Huang y
Yan Zhao
Yan Zhao