An Association Rule-Based Multiresource Mining Method for MOOC Teaching.

The selection of MOOC teaching resources is influenced by diversified resource positioning methods, which leads to low index efficiency of resource mining. Therefore, this paper proposes a multiresource mining method based on association rules to collect the learning behavior data of MOOC users and establish the MOOC teaching resource warehouse. Aiming at the attribute set of information association positioning, the association rules of teaching resources are designed. In addition, the association rules are combined with the shortest path scheduling scheme of teaching resources to establish the location and mining of diversified MOOC teaching-associated resources. Finally, the clustering method is used to process the results of teaching resource mining and complete the clustering of diversified teaching resources. Experimental results show that the index time required by the proposed mining method is 0.1 s, which is only 1/6 of other resource mining methods.

1. Introduction

Driven by the internet, MOOC emerges as a large-scale open online teaching model. The use of the internet and modern communication technology to collect useful teaching resources from all over the world, with the help of the internet for real-time dissemination, ensures that every learner can access resources and learn effectively anytime and anywhere [1]. As education evolves with the times, MOOCs have emerged as a new teaching tool. English instruction may benefit from its benefits of liberalization, diversity, and real time, which are worth additional investigation and careful examination. There are no time or place restrictions for MOOC learning, and the method of instruction is adaptable, making it ideal for pre- and postassessment preparation [2]. Short and thorough video material and unique modularity characterize MOOCs in essence. According to a question, knowledge points are arranged and then instructional activities are carried out. A typical lecture by a teacher lasts anything from five to fifteen minutes. Fragmentation is also seen in the instructional material of instructors. It is important to break down complex concepts into a series of small knowledge points that can be taught step-by-step with the use of films and animations [3]. A vast number of teaching materials are needed for Chinese students to learn English in MOOCs since English is a foreign language [4]. There are a lot of different sorts of instructional materials that have to be managed, and they are all time-consuming and complicated [5], including media material database, question database, case database, curriculum database, resource index database, and network course database, which urgently need to achieve scientific information management. Based on the teaching background of MOOC and combined with the actual teaching needs of higher vocational English, this paper focuses on exploring the diversified mining methods of English MOOC teaching resources to realize the development and management of teaching resources [6].

In the age of the knowledge economy, the online teaching resource library, as a digital teaching resource center, is a key component of creative education and training of college students with professional knowledge and information literacy [7]. The resource mining approach proposed in this study considerably simplifies data integration and transformation by using association rules. Through unbiased and objective statistics and analysis [8], it can effectively organize complicated data, rapidly and properly uncover hidden patterns, and precisely comprehend future dynamics, as well as increase the efficiency and ability of data use. Given the complexity of the data created throughout the English MOOC education process, typical statistical analysis approaches are ineffective in revealing hidden knowledge and principles. Data mining technology is employed to examine the online learning data in this article. The resource diversification mining approach described in this study successfully enhances the data index efficiency of resource mining, according to experimental findings [9].

2. Design of Multiresource Mining Method for MOOC Teaching Based on Association Rules

2.1. Designing the English MOOC Teaching Resource Warehouse

Users' learning behavior data is in a variety of forms, including structured and semistructured information, due to the complexity of MOOC teaching situations [10]. The server data side, client data side, and proxy server data side are usual where this information is maintained. An English MOOC teaching resource warehouse is built in this study in order to establish suitable conditions for data mining, and the course information included in English MOOC is gathered as the data foundation for the resource warehouse's creation [11]. Table 1 summarizes information on English MOOCs.

Table 1

Data set field information.

Project	The specific content
Information provided by the user	User ID
	Record of formal schooling
	Age
	Gender
The system records learning behavior	The system records learning behavior course number
	Forum usage times
	Whether the browse
	Interaction of days
	Video playback
	Withdrawal date
	Number of interactions

Based on the aforementioned gathered data, the creation of an object-oriented, integrated, nonvolatile, and time-variable data warehouse is beneficial to management decision support [12]. The database system should be compatible with current computer technologies. To demonstrate the internal connection and operating mechanism of the teaching process under certain circumstances, the teaching material should be correct, and the teaching method should be scientific and reasonable [13]. The system's usability is the second consideration. The database system support for teaching application is simple and flexible [14], and it can meet the needs of different ways, with a wide range of applications, simple operation, complete disciplines, rich content, and diverse forms, and it can meet the needs of classroom teaching of all grades and levels, as well as the needs of individualized and personalized learning. The data warehouse can firstly be considered as a special D8M5, as shown in Figure 1.

Figure 1

Structure diagram of English MOOC teaching resource warehouse.

The process of constructing a teaching resource data warehouse entails switching from a standard database-centered operational system structure to a data warehouse-centered system structure [15]. The first step is to create an architectural change by determining what data is existing in the system, which serves as the foundation for creating a data warehouse. Finally, for instructional resource information, characteristic variables are discretized. The number of workers included in the discretization value of each feature after discretization is near to the criteria of discretization, and the discretization process is indifferent to the discretization dividing points [16]. The data are discretized for the distribution drawing of the numerical feature values according to these criteria, which are based on the valley bottom classification of the graph in the data classification summary. You may choose the data mining system theme based on the study of user demands once you grasp the data that the system has. Many business systems are involved in a big data warehouse system, and their roles are complicated. As a result, the spiral development approach is used to divide and conquer the massive objective by breaking it down into various phases of execution. The subsystem of mining association rules is one of the simple and clear problems.

2.2. Designing Association Rules for Teaching Resources

The educational process must follow the necessary educational conduct regulations, and educational resource mining must also follow specific guidelines. The article should be based on resource mining ideas before establishing the association rules for teaching resources [17]. We must pay attention to whether the resources have educational significance throughout the entire process of mining ecological values education resources, in order to fundamentally grasp the direction of resource mining, clear mining content, and grasp the development law of college students' values cultivation. Simultaneously, the diggers must adhere to national policies, comprehend the Ministry of Education's and national leaders' requirements for the development of ecological values in college students, adjust the direction and scope of mining in a timely manner, and dig out potential educational resources to meet the needs of English talent cultivation.

In order to better define the field and focus of resource mining, it is necessary to combine the current situation of current college students' ecological values acquisition with the new requirements put forward by social progress for the majority of young students in the process of mining ecological values education resources [18]. To achieve enhanced mining education resources, start as far as possible from the current state of environmental conditions, the education object, targeted mining, composition and configuration of education resources, and for different levels of education object to select appropriate teaching resources, to promote the effective use of education resources and to avoid all types of idle and waste of resources. In addition, when mining educational resources, we should not only expand our vision to relevant policies, laws, and regulations at the national level but also focus on the development and changes of the real-life environment. Starting with life, pay attention to uncovering and mining MOOC teaching materials in life, bringing education resources from life, and integrating education into life. Because of the nature of life, college students will be more likely to grasp and accept the contents of the education if they are given some vivid examples in real life. This will allow them to practice English in real life. It is vital to secure the systematicness, integrity, and coordination of resources in the process of resource mining for ecological values education. Because everything is connected, we should apply the rule of universal connection while mining instructional materials. To improve learners' English quality, a series of selection activities must be carried out for both explicit and implicit resources, such as reasonable allocation, selection, screening, addition, and combination of various resources, in order to make educational resources a reasonable, controllable, orderly, and moderately effective resource [19]. It is impossible for any educational resource to carry out educational activities alone and obtain optimum educational outcomes in teaching activities. Only when all types of resources are combined and coordinated to create a system and a whole organism can they interact and cause consequences. Furthermore, the combined benefits of educational resources will much outweigh the effects of individual resource pieces.

Based on the above principles, an association rule is established for the attribute set of information association positioning in MOOC teaching resource warehouse. Concept lattice and association rule attributes and the reduced concept lattice between them are used to generate frequent item sets of information association positioning and extract implication rules. The support vector set is a subset of interesting rules in the data mining transaction set, and the concept lattice on FP-tree is used to represent the support degree of the two sets, so as to obtain the probability set of redundant association rules, and the calculation formula is as follows:

In formula (1), X and Y represent the attribute set, S represents the support between two attribute sets, and P represents the probability of redundant association rules. Considering that the associated data of MOOC teaching resource information has the characteristics of regular state distribution, the probability formula of distribution condition is as follows:

In formula (2), C represents confidence degree, and the characteristic scale of associated data of MOOC teaching resources under association rule training is called minimum confidence minconf. When the support of associated data is greater than the given minimum support minsup, the synthesis rule in the joint probability dense function set D is a nonempty subset, called frequent item set. Using frequent item sets to design association rules, the closed frequent item mathematical model of mining associated data of MOOC teaching resources is as follows:

In formula (3),

In formula (4), W represents closed frequent item set, l represents the label of mining node, k represents the feature vector of teaching resource, η represents the test time window of data, N represents the data scale, x represents the constraint condition, and r represents the constraint set. Through the adaptive configuration of the fitness of the difference features and under the guidance of the synthesis rule of associated data mining, the association rules of association positioning are designed as follows:

And get the formula as follows:

In formulas (5) and (6), m represents associated data, n represents the total number of training samples, A represents constraint concept lattice, and K represents association rule. Pheromone guiding of information association location is carried out by obtaining training samples of network education information association data from the information database and developing the association rules of information association location.

2.3. Positioning MOOC Teaching-Related Resources

According to the above information association positioning rules of MOOC teaching resources, the improved design of information association positioning mining method is carried out. This paper proposes a positioning mining method for MOOC teaching-associated resources based on shortest path scheduling of educational information resources. The shortest path scheduling of educational information resources adopts the constraint concept lattice interval scheduling model, and the constraint set of shortest path relationship of educational information resources is as follows:

In formula (7), the minimum support threshold constraint method is used to track the training process of data mining. Associated resource mining of MOOC teaching resources is a process of constraint evolution, which can be expressed as the following optimization formula:

In formula (8), J represents the constrained evolution model, and e represents the number of mining training. The teaching source shortest route scheduling is done using a restricted idea lattice and the best combination of candidate nodes. The instructional resource diversity selection gene then defines the shortest route scheduling procedure on the limited data set at the frequent nodes that match the conditions, given the degree of support.

The constraint requirements of network education resource shortest route scheduling are established using the extracted constraint association rule database and the support threshold of frequent item set. The restricted idea lattice Hasse diagram of MOOC teaching resource association placement is produced based on the given requirements. The frequent item set characteristics of every object information in the network education information system are retrieved using the shortest route scheduling results of educational information resources. The priority attribute list of association rules of the network education information system sample database is provided in Table 2 under the restriction of minimum anticipated support number.

Table 2

Children's preference for sports to enhance physical health and its reasons.

The serial number	Association rule item set	Priority count
1	AT-tree	2
2	Root	4
3	Link	1
4	AP	5
5	E	3

The core of the association rule mining algorithm is to find frequently occurring item sets, which determines the algorithm's overall performance. Multiple database scans are frequently required to solve this problem, implying that a significant amount of time will be spent on database scans and IIO operations. As a result, the main problem that all types of association rule mining algorithms must solve is how to find all frequent item sets quickly and efficiently, and it is also the standard by which to compare the benefits and drawbacks of various association rule mining algorithms. The frequent pattern tree of information association location mining is constructed as shown in Figure 2 using the priority attribute list of association rules shown in Table 2.

Figure 2

Frequent pattern tree for resource mining.

The frequent item set features of diversified object information in the teaching resource warehouse are extracted from the frequent pattern tree, and the node of concept cell is expressed as C((A, B)P), whereas, Р is the infrequent item in the transaction item set, A is the minimum expected support number, and B is the correlation positioning connotation after the transaction item set is processed. The abovementioned constraint criteria are used to carry out the information association location mining. The input initialization parameter value, the new head table, and the constraint index parameter set of data mining are all input when the resource diversification mining method is used. Obtain information about an association's location by mining the two-dimensional table's predicted support number. Frequent patterns are mined in the global tree according to the new head table, and the result is a two-dimensional table form with 0, 1 values, where 0 indicates that the mined information data does not meet the minimum expected support set and 1 indicates that the mined location information results meet the minimum expected support set. The diverse mining results of MOOC teaching materials are achieved using the following mining approaches.

2.4. Realizing the Clustering of Diversified Teaching Resources

One of the essential technologies of the association rule mining technique described in this research is whether the data can be successfully clustered, and one of the core difficulties of clustering is the selection of the correlation function. The clustering approach is used to deal with items in categorised data after mining instructional resources based on association criteria. A hierarchical clustering approach based on item correlation function is designed to suit the clustering needs of diverse instructional materials. Starting clustering from the items of the first or final layer (leaf layer) of the classification data to the higher or first layer of the leaf layer, a top-down or bottom-up technique is used. The minimal support for each layer of clustering operation is lowering in the hierarchical clustering process of the items at the first layer. The following is the particular operation:

To begin, each item in the frequently used item set is separated into classes. To decrease computational complexity, the classes with the highest similarity are joined using the correlation between them. Finally, using the concept of hierarchical clustering, the two classes with the highest degree of similarity are explored further to see whether they may be combined. The first-level item clustering of MOOC instructional materials is complete at this stage. The same resource clustering strategy is utilized for successive clustering after the first clustering. Clustering categorised data pieces has the primary goal of reducing and dividing the database. This article proposes a database reduction strategy that is basically a specific “partitioning” method. The reduced method, as opposed to the traditional method of dividing transaction databases by transaction records, divides the items in transaction records, allowing high-correlation items to be classified into the same transaction database and diversified clustering of teaching resources to be realized.

Finally, the AFOPT-tree compression structure used in this study is appropriate for varied clustering of instructional materials, and the size of the AFOPT-tree is further reduced by subtree merging. The AFOPT-tree is further lowered when each resource item is examined by merging its relevant subtrees. After establishing all frequent item sets with the prefix of the item created in each inspection, the node X corresponding to the item generated may be pruned from AFOPT-tree, and a merger operation of the node idea tree can be added using the top-down and depth-first traversal approach. The so-called subtree merging procedure involves comparing all child nodes of X node with all sibling nodes of X node, merging the node if an item with the same name exists, and adding up the node count. If this is not the case, a new child node is placed as the parent node of X node. In general, even though this step adds overhead, it is still preferable to enhance the algorithm's efficiency after AFOPT-tree compression. In addition, two auxiliary data structures, a head table and a subhead table of items, are introduced in this study. Based on the AFOPT-tree structure, all frequent item sets with X as the prefix are generated. Subhead table pointers may be used directly to identify nodes in the AFOPT-tree, allowing the original AFOPT-tree to be reused. Reconstructing the conditional sub-AFOPT-tree adds unnecessary overhead. Filtering pruning is accomplished using classification features, and filtering pruning optimization strategies are also presented based on classification data characteristics. If an extension item set has an ancestor or descendant connection with item X, it is not entered into the child header of item X, and the search for following extension items is not resumed. The optimization rule can effectively reduce the candidate item set size of all frequent item sets prefixed with this item and improve the clustering efficiency of diversified educational resources. By clustering diversified teaching resources through the above operations, diversified mining of English MOOC teaching resources based on association rules is realized and the MOOC teaching resources are enriched.

3. The Experimental

Experiments are carried out using this method; the experimental operation is carried out, so as to clarify the superiority of the resource mining method designed in this paper.

3.1. The Experimental Data

In order to enhance the authenticity of experimental results, an English MOOC teaching software was selected to collect experimental data. The MOOC software has 10TB of English teaching resources, with more than 4,500 registered users and more than 300 visitors per day. Based on the MOOC software, the experimental data are obtained as shown in Table 3.

Table 3

Experimental data set.

Project	The numerical
Number of users/person	400
English MOOC teaching resources/per	1400
User comments/item	32000
The level of data set sparsity	0.94

Based on the above experimental data, it is divided into four groups of different numbers of data sets. The mining method designed in this paper, BP neural network mining method, decision tree mining method, and particle swarm mining method are used to conduct the simulation experiment of diversified mining of MOOC teaching resources.

3.2. Experimental Simulation Results

Simulation experiments were conducted according to different resource mining methods. Considering the differences of English MOOC teaching courses, the class clusters were divided into 3 to obtain the teaching resource mining results shown in Figure 3.

Figure 3

Different methods of teaching resource mining results.

According to Figure 3, BP neural network mining method, decision tree mining method, and particle swarm mining method have similar resource diversification mining results. Among them, there are some mixed resources in BP neural network mining results, and the amount of teaching resources mined is small. There is no mixed data in the teaching resources obtained after the application of decision tree mining method, but the teaching resource mining results have little basis. Compared with the previous two mining results, the particle swarm mining method shown in Figure 3(c) effectively improves the amount of resource mining. However, the application of the design method in this paper realizes the mining of high-density teaching resources. Starting from each MOOC teaching resource, diversified teaching resources are mined outward to obtain a large number of high-precision teaching resources. To sum up, the teaching resources obtained by the resource mining method designed in this paper are far superior to other methods, which proves the effectiveness of the design method in this paper.

The experimental data were directed to a grid that was separated into horizontal and vertical coordinates, with the vertical representing the quantity of data and the horizontal representing the time. Figure 4 depicts the correlation detection findings of mining result resources using the design strategy used in this work.

Figure 4

Mining resource quantity curve.

According to Figure 4, after the application of the mining method designed in this paper, the correlation detection effect changes more and more significantly over time. When the experiment time is 16 s, the maximum value of database resources is 7450, indicating that the improved technology has a higher data correlation.

3.3. Index Time Comparison

In order to clarify the advantages of the design method in this paper, before the software is used to complete resource mining, the time function is added in the front and the back of the program, respectively, and the time difference obtained is the time of the method to execute the class. Through the above operations, the index time of different mining methods is calculated and obtained.

Set the number of index words to 3 to analyze the index efficiency of four different resource mining methods. According to Table 4, after the application of the design method in this paper, the index time is 0.1 s, while the index time required by BP neural network mining method, decision tree mining method, and particle swarm mining method is 0.3 s, 0.5 s, and 0.8 s, respectively. By comparing the experimental results of the data mining technology of four different data mining methods, it is obvious that the data index efficiency of the mining technology designed in this paper is the highest, which is nearly one-sixth of the index time of the other three data mining technologies. The index efficiency is greatly improved, and the experimental results verify the practicability of the improved technology.

Table 4

Index time comparison.

Method	Number of index words/per	The index of time
The mining method is designed in this paper	3	0.1
BP neural network mining method	3	0.6
Decision tree mining method	3	0.7
Particle swarm mining method	3	0.8

4. Conclusion

Aiming at the problem of low mining efficiency of traditional methods, this paper proposes a mining method of information association location based on association rules, applies association rule mining to English MOOC teaching resource mining, analyzes the association relationship between resource information, and realizes efficient resource mining. In this paper, a method of English MOOC teaching resource information association location mining is proposed, and the performance of information location mining is improved through the optimization algorithm design of the location mining method. The results show that the index efficiency of information association location mining in online education information system by using this method is greatly improved, and it has good application reliability, which shows the good application value of the method presented in this paper. Although the multimining method of teaching resources in this design has carried out a large number of experiments and engineering usability research and has made some achievements and progress, there are still some problems. The following are the author's suggestions for the next research work; only by using heuristic conditions and analysis of hyperlink anchor text can we obtain the links of teaching resources and recycle the web pages to expand the amount of teaching resource collection.