Material selection system of literature and art multimedia courseware based on data analysis algorithm

2.2.1 DM

DM is a collection of knowledge from multiple disciplines, and its definitions are also varied. When it comes to DM, it is associated with the term Knowledge Discovery in Database (KDD) in the database. For a long time, there have been different opinions in the academic community on whether KDD and DM are a proposition. Many scholars consider the two to be synonymous, but with different names. Therefore, DM literally means “uncovering knowledge from data.” In addition, some scholars believe that DM is an important part of KDD. KDD focuses on understanding the knowledge in the entire database, and DM is the core of the entire process, which can find hidden models for evaluation through DM, so its scope is much larger than data mining [11,12]. The specific situation is shown in Figure 2.

Figure 2

Knowledge discovery process.

Data cleaning: various data are filtered out to remove noise. Data integration: various data sources are integrated. Data selection: data relevant to the target is extracted from the data set. Data transformation: data is transformed into data suitable for mining operations through specific operations. DM: data types are discovered and selected. Model evaluation: useful and meaningful knowledge for users is obtained according to an evaluation method. Knowledge representation: existing visualization techniques are applied to existing knowledge acquisition and application [13,14].

DM can help uncover underlying patterns in huge data sets that may not be easily accessible by human analysis. By mining hidden patterns in data, it is possible to gain insights into business, market or customer behavior, and make more informed decisions. DM relies on high-quality data, but the real data often has problems such as noise, missing values and errors. If the data quality is not high, this can lead to inaccurate or misleading mining results. The overall purpose of DM is to extract relevant information from a set of information and associate them with an integrated structure for use in future applications. In DM, classification is a very important mining algorithm, which is a process of finding classifiers [15]. It analyzes objects within a dataset using specified class labels, typically using a training set in which all objects are associated with known class labels [16]. The method first learns and constructs a model from the training set, and then uses the model to classify new objects. In other words, it can be said that classification is the grouping of data into categories and is an increasingly popular method of dealing with a wider range of data.

Prediction accuracy is a commonly used index to evaluate the effect of classification algorithms. It measures the accuracy of classification algorithms in forecasting. Prediction accuracy refers to the ratio of the number of samples correctly classified by the classification algorithm to the total number of samples on a given test data set. The prediction accuracy ranges from 0 to 1 and can be converted to a percentage representation. For example, a prediction accuracy of 0.85 indicates that the algorithm has 85% classification accuracy on the test data set.

The high prediction accuracy indicates that the classification algorithm has good prediction ability on the given test data set, that is, it can accurately classify the samples into their respective categories. However, it should be noted that the prediction accuracy only reflects the performance of the model on the test data set, and does not fully represent its performance in practical applications. Whether the quality of a classification algorithm can meet the needs of the designer can be measured by several indicators in Figure 3.

Figure 3

Metrics to measure.

2.2.2 DT classification algorithm

DT classification is a supervised classification method. The method can handle massive amounts of data efficiently and does not require much-specialized knowledge to build the tree. It is often used for exploration, and the result is the knowledge gained from the tree, which is very intuitive and easy to accept. DT is a very representative classification algorithm. The algorithm uses a greedy method and a top-down, non-reverse approach to decompose a training set into several smaller subsets. Finally, an intuitive tree is formed, which makes the characteristics of the training set more clear [17,18].

2.3.1 Basic algorithm

ID3 is the earliest DT classification method in the world. It is based on the “information entropy” of ID3, which is the C4.5 algorithm. Therefore, for explaining the algorithm of C4.5, the first thing to do is the introduction of ID3.

The basic idea of ID3 algorithm is: by analyzing the information gain of the attributes, the optimal segmentation attributes with recognition performance are found, and the sample set is divided into several subsets; then, similar recursive decomposition is performed on each subset, and finally a DT is obtained [19].

From the basic idea of the algorithm, it can be seen that how correctly determining the most valuable division attributes is the key to realizing this method. In response to this end, entropy in information theory is cleverly used to measure the gain of information.

Entropy is a measure of the burstiness, uncertainty, or randomness of a data set. For given possibility a 1 , a 2 , … , a d , there is:

then entropy J ( a 1 , a 2 , … , a d ) is defined as:

The entropy value is 0–1, and the entropy is 0. a o means that all the data in this set are of the same type. If all the data in the set are equally distributed, then the probability of each data is the same, and the entropy value is the largest.

It is assumed that a database is well-ordered, that is, without heterogeneous tuples, then no further segmentation is necessary. In addition, each time it is swiped, a maximum split state is selected. The information gain calculation method is: by comparing the sum of the weights of the entropy of the original data set and the entropy of the divided data set, the weight value is expressed by the proportion of the decomposed subsets in the whole data set [20].

The literature and art multimedia courseware materials are tried to be classified. It is supposed that D is a set composed of d courseware materials, which consists of m different material type attributes V o . Among them, o = 1 , 2 , … , n , the number of materials in material type V o is D o ; then, the amount of material information required to classify courseware material D is:

Among them, a o represents the proportion of the number of courseware materials D o in the total samples in the material category V, and the calculation is:

It is assumed that a courseware material attribute S has m different values { s 1 , s 2 , … , s m } , and by using the courseware material type attribute S, the sample set D can be divided into m different material sets { D 1 , D 2 , … , D m } . The data sample when the material attribute S is s k is included in D k . If the material attribute S is marked as a divided material attribute, that is, the current courseware material is divided into S, then d o k is used to represent the quantity of the visible material D k belonging to category V o . At this time, the calculation of the material type information is:

Among them, d 1 k , d 2 k , … , d n k d represents the weight of the k subsets D k . The smaller the value obtained by R ( S ) , the more thorough the classification of the subset and the higher the purity. In this case, the calculation of the information amount of D k material is:

In these data, a o k represents the proportion of data samples of subset D in material category V o . Thus, the material type information gain is obtained from the set of molecules delineated by the material attribute S:

According to formula (7), the material information gain of each attribute can be compared to obtain the largest attribute, that is, the best attribute. It is used as the split attribute in the sample set; then, all the data are put into the same category, and finally, only one DT is left.

2.3.2 C4.5 algorithm

The ID3 algorithm has great limitations. In response to this problem, the researchers put forward a more perfect C4.5 algorithm, the core of which is “information entropy.” It is optimized and improved on the basis of ID3, and it guarantees the accuracy and speed of classification [21,22]. Based on the algorithm idea of ID3, C4.5 has improved the algorithm based on ID3:

If the information gain ratio is used as a new feature recognition ability index, the problems existing in the ID3 algorithm can be well solved.

The material information gain rate refers to the ratio of material information to segmentation information, and its expression is:

The utilization of material information gain rate can solve this problem, but it also brings a new problem: if it is a ratio, then the denominator would become 0 or very small (when something is close to D), so the result is that the ratio is very high or undefined. In order to prevent this, the gain calculation of material information is divided into two steps: the first is to calculate the gain of the material information, which ignores the properties whose results are lower than the average, and only calculates the gain of the data. Then, according to the gain coefficient of the data, the best segmentation attribute is selected [23,24].

2.3.3 Improvement of C4.5 algorithm

It is assumed that there are only positive example set D F and negative example set M F in the class attribute of the courseware material data set F, and their sizes are u, m respectively. It is assumed that in the vector space R, the probability of a correct DT classifying any set of data matches the positive and negative probabilities of R. By combining with formula (2), it can be known that the amount of material type information required by a DT is:

The material attribute C is selected as the root node of the DT, and C has D different material values { C 1 , C 1 , … , C d } . By using the material attribute C, the courseware material data set F can be divided into d subsets { F 1 , F 1 , … , F d } , wherein F o contains the courseware material sample data whose attribute C value is C o in F. Similarly, it is assumed that F o contains u o positive example sets and m o negative example sets, then the expected material information calculated by subset F o is O ( F o ) . From this, it can be concluded that:

Therefore, it can be concluded that the material information entropy required to take C as the root node of F is:

After simplification, there is:

Since 1 ( u + m ) ln 2 is a constant in the training set, its calculation process can be simplified. Therefore, it can be assumed that:

According to the mathematically equivalent infinitesimal principle, when c is very small, then ln ( 1 + c ) = c . From this, it can be concluded that:

After combining formulas (9), (14), and (15), it can be obtained:

After combining formulas (13)–(15), it can be obtained:

The material information gain rate is:

On this basis, a new method ∑ o = 1 b 2 u o m o u o + m o based on information entropy of courseware material classification is proposed. This method uses the expression 1 ln 2 u o m o ( u o + m o ) 2 of the material information entropy to calculate the split information amount of the material. In the calculation process, constants do not need to be considered, but the attribute with the largest gain coefficient of the material data is used as the measurement standard. The new method only contains four mixing operations compared with the previous method; thus, the computational efficiency is improved [25].

3.1.1 Questionnaire design

Through the analysis and research of theories and examples of multimedia courseware in colleges and universities, and combined with the relevant literature, the relevant questionnaires were compiled. From April 2022 to June 2022, a total of 300 questionnaires were distributed to teachers and students of three universities in Z city. Among them, 80 questionnaires were issued to teachers of literature and art, and 80 questionnaires were recovered. There were 220 questionnaires sent out to students, and 200 available questionnaires were returned. All returned valid questionnaires. After the survey was completed, EXCEL tables were used for statistics and analysis of the survey results.

3.1.2 Questionnaire survey of teachers

Figure 4 shows some data on teachers’ use of multimedia courseware.

Figure 4

Teachers’ usage of multimedia courseware: (a) Recognition of the importance of multimedia courseware, (b) source of multimedia courseware, (c) time spent in making courseware, and (d) survey on the number of pages used in multimedia courseware.

It can be seen from Figure 4(a) that among the teachers surveyed, 61.25% of the teachers rated it very high; 32.50% of the teachers thought it was generally important, and 6.25% of the teachers thought it was possible to have it or not. It can be seen from this point that most teachers felt that the use of multimedia courseware was very important in literature and art classes.

It can be seen from Figure 4(b) that 22.50% of teachers’ courseware was written by themselves, 21.25% of teachers were downloaded from the school’s resource library, 56.25% of the teachers downloaded from the Internet, and none of them were bought from the publishing house. It can be seen that the current multimedia courseware was mainly the ready-made courseware downloaded from the school resource library or the Internet, and few teachers would make their own according to the needs of the classroom.

From Figure 4(c), it can be seen that in terms of production time, 58.75% of the teachers who participated in the test chose no more than 2 h; 41.25% of teachers chose 2–4 h, and no one chose 4–6 h or 6 h. It can be seen from the analysis that teachers spent more time in the process of making multimedia courseware.

From Figure 4(d), it can be seen that for the courseware used in a class, 13.75% of the subjects chose pages 10–20, and 68.75% of the teachers chose pages 10–20; 17.50% of the teachers chose 20–30 pages, and no one chose 30–40 pages.

It can be seen from Table 1 that in the survey on the reproduction ratio of courseware content, 36.25% of teachers chose more than 2/3, and 53.75% of teachers chose 1/2–2/3; 10.00% of teachers chose 1/3–1/2, and none of the teachers chose less than 1/3. Therefore, it can be considered that teachers of literature and art used a large number of multimedia courseware in the classroom, and the main content of these courseware was to display the content of the text.

From Table 2, it can be seen that in the survey of the proportion of materials, 62.50% of teachers mainly focused on writing, and they emphasized the key points and difficulties in teaching; 26.25% of teachers used pictures as the main content to increase students’ visual experience and create a situation; 11.25% of teachers chose video as the main method to activate the classroom atmosphere. According to these materials, teachers’ multimedia teaching content was mainly text and supplemented by appropriate pictures and videos.

From Figure 5, it can be seen that in the survey on the advantages of multimedia courseware, 32.50% of the teachers surveyed believed that this could enrich the teaching content and make the classroom atmosphere more vivid; the survey results showed that 42.50% of teachers believed that this can effectively enrich the teaching content and improve the teaching effect; 21.25% of teachers believed that writing on the blackboard can save a lot of time, and it was more convenient; 78.75% of teachers believed that it can make students immersive; 93.75% of the teachers thought that it can display the teaching content more vividly and highlight the difficulties of teaching, and 63.75% of the teachers thought that it can make classroom teaching more interesting and arouse the interest of students. Through these analyses, it can be seen that teachers had a deeper understanding of the accessibility of multimedia courseware and had realized and experienced many of its benefits.

Figure 5

Advantages of multimedia courseware (teacher).

From Table 3, it can be seen that while learning the benefits of multimedia courseware, teachers also found some problems. 83.75% of the subjects said that the use of multimedia courseware in the classroom would affect the enthusiasm of students; 33.75% of the teachers thought that the courseware would aggravate the teacher’s inertia; 57.50% of the teachers thought that if the production was too delicate, it would distract the students; 78.75% of the teachers felt that the use of multimedia courseware in the classroom made the teaching process rigid and the flexibility decreases, and they cannot operate flexibly according to the feedback of students; 47.50% of teachers believed that the excessive use of courseware made the interaction time between teachers and students short and cannot provide timely feedback. From the survey results, teachers had a big problem in the use of multimedia courseware, that is, the use of courseware was too rigid, which affected the subjectivity of students.

3.1.3 Student questionnaire

The purpose of the research is to understand students’ subjective willingness to courseware, and to understand their views and opinions on teachers’ use of courseware.

From Table 4, it can be seen that 79.00% of the 200 students felt that they can understand better and more deeply through multimedia courseware; 21.00% of people felt that the knowledge understanding in traditional classrooms and multimedia classrooms was the same, and no one felt that multimedia teaching was inferior to traditional classrooms. It can be seen from this point that students had a positive attitude towards the effect of multimedia teaching.

Figure 6 shows some views of students on multimedia courseware. From Figure 6(a), it can be seen that the thinking time given by literature and art teachers when using multimedia courseware in class is as follows: 23.50% of the people thought that the thinking time was relatively long; 44.50% of the people thought that the thinking time was average; 32.00% of the people thought that the thinking time was too short. According to the above information, most students felt that when teachers used multimedia courseware in class, they did not give students enough time to think, which affected their thinking ability.

Figure 6

Some views of students on multimedia courseware: (a) thinking time given by teachers in multimedia classrooms, (b) survey of preference for courseware materials.

It can be seen from Figure 6(b) that 14.00% of the students liked words, and 26.00% of students liked pictures and diagrams; 59.00% liked animation and video, and 40.00% liked music; 52.50% liked to have interaction. From this, it can be seen that students’ preferences for materials were different. Therefore, the different needs of students should be taken into account when making courseware, and the diversity of materials should be achieved.

It can be seen from Table 5 that among the 200 students, 19.00% preferred to have one-on-one and face-to-face communication with teachers, and 60.50% of students were willing to spend more time in class on thinking, questioning, and mental preparation; 20.50% of the students liked to listen to what the teacher talks about, and 20.50% of the students liked to listen to what the teacher talks about. It can be seen that students are expected to communicate more with teachers and make full use of their own subjectivity.

From Table 6, it can be seen that students had different views on the superiority of multimedia courseware. 64.00% of the students believed that the courseware could better display the teaching content, and 70.00% of the students believed that the forms were diverse; 62.00% of the students felt that the courseware could effectively improve the teaching efficiency, and 55.00% of the students felt that the courseware could allow teachers to save time and provide more knowledge on the blackboard; 40.50% of the students feel that the courseware can stimulate their thinking about the problem.

From Table 7, it can be seen that with the popularization of courseware, students had realized the problems brought by the use of courseware. About 4.00% of the students felt that the multimedia courseware made their study time too long and could not be digested in time; 48.50% of the students felt that the time for taking notes when using the multimedia courseware in the classroom was too tight, and there was not enough time to listen to the teacher’s lectures; 22.00% of the students felt that the problem was that they were too far away from the computer screen to clearly see the content of the courseware; 72.00% of the students felt that the interaction time with the teacher in the classroom was too short; 36.50% of the students believed that the light on the multimedia display was too strong and too dazzling.

3.2.1 Overall system architecture

The selection and sharing of multimedia courseware materials is a greater advantage to realize network teaching, and it is also in line with the needs of the development of educational informatization. Through the integration of existing teaching resources and the construction of a multimedia resource library, a large number of high-quality teaching materials can be brought together to realize the sharing of resources [26]. In addition, users of the system can also make full use of supplementary recording and uploading to make up for the deficiencies of the system, which provides more abundant resources for the majority of users.

When establishing a multimedia data management and sharing system, the following points should be noted: system users can obtain authorized data through the Internet at any time under their authorization; the information architecture of the system is good, and services can be invoked through standardized interfaces to ensure the safe transmission of data; the management of material resources requires massive amounts of data and requires decentralized management in the massive amounts of data [27]. Figure 7 is the overall structure of the system.

Figure 7

Overall architecture diagram of the system.

According to the needs of the system itself, it is necessary to solve complex resource management problems and to give full play to the role of business logic to meet customer service requirements for the system. In order to facilitate the user’s access, the B/S structure is adopted, and it is designed and developed. In Figure 7, the system is divided into user layer, application layer, resource layer, and network layer. It adopts a hierarchical structure to realize the separation of business logic and technical logic, with a clear structure and strong maintenance ability [28,29].

3.2.2 System software architecture

According to Figure 8, the software architecture is divided into several parts: background program, database, control engine, permission rules, and users.

Figure 8

System software architecture.

Background program: the background program is used to process the business and data of the system, and input the material resource data into the database.

User interaction interface: the user submits service processing requests to the system within the scope of authorization, and can control the service process and call functional functions.

Database: asset information, user information, etc., are stored in the database.

In the entire architecture, the core of the entire system architecture is to coordinate and control data processing, business operations, and communicate with users through the connection between the control engine and other components. The main function design of the control engine consists of two parts: initiating the central controller and front-end request processing.

3.2.3 Design of data entry system

The main research contents are the input and sharing of multimedia courseware materials. The input of data into the system mainly includes two steps: the first is to go through a cataloging process to become the original material; the second is to use format conversion, secondary cataloging, etc., to convert them into target materials, and set some materials as shared materials as needed. According to the main process of data input, the data input system can be divided into two modules: front-end resource storage system and material resource management platform. Resource Manager is a system resource manager that provides computing capabilities and resource pools for users by configuring and scheduling computing resources. Its functions include resource management, life cycle management, and task management. Audio Video Interleaved, that is, audio video interleaved format. Figure 9 is a system architecture diagram of the front-end resource storage and source management part.

Figure 9

System architecture diagram of front-end resource storage and resource management part.

3.2.4 Shared resource library design

The construction of multimedia material resource management and sharing system is to achieve the purpose of resource sharing through input, conversion, and management of multimedia material resources. Through the establishment of this system, the sharing of multimedia material resources among different colleges can be realized. Through the resource-sharing network, colleges and universities have realized the sharing of multimedia material resources and the sharing of school resources. The following is a detailed introduction to the resource-sharing mechanism of each college. The system uses a distributed resource storage framework with centralized index storage. Figure 10 is a diagram of a centralized index storage structure.

Figure 10

Centralized index storage structure diagram.

In Figure 10, the system uses a centralized index storage structure with the school as the central server. Under the traditional centralized sharing architecture, the central server is responsible for saving the content index of the shared nodes. When a shared node client sends a content request to the central server, the central server queries the stored index and returns the stored physical address of the server to the client. However, the central server of the system has two functions of index service and data service resource at the same time.

Central server: the central server has two functions of data service and index service.

Indexing service: the central server saves the resource tables of the resource libraries of each college. When the clients of each college send a query request to the central server, the server obtains the form and returns the school server address with the target resource to the requesting client.

Data service: part of the shared multimedia data is stored in the central server, which can be accessed directly.

School server: each college server is a connection node of the TCP/IP network, as shown in Figure 10. The school’s servers are divided into two types, one is shared resources and the other is private resources, which can only be accessed by students of the school. The communication between the central server and each college is bidirectional. The central server obtains the information shared by each college from the server of each college, and when the user of the college asks the central server, the server entity address of the target resource would be transmitted to the client; at the same time, the sharing of network resources can be realized through the network communication among various universities.