Research on Physical Education Teachers’ Role Change and Teaching Innovation Practices in Colleges and Universities in the Digital Era

Prepare the video resources of the teaching practice course for 2-3 weeks in advance before the class, which mainly consists of 2-3 micro-videos, the content of which mainly includes the overall structure of a completed technical action, decomposition of the action, the key points and the difficult points, and record it as a completed technical action flow system. After successful recording, the video material is uploaded to the WhatsApp platform, through which students can learn in a targeted way. Under the conditions of their own actual situation, they can grasp the progress and rhythm of learning.

In terms of targeted practice and communication, the practice tasks should be combined with the teaching content in the micro-video to strengthen the consolidation and improvement of knowledge points, and the students can also discuss and interact with the teachers and students through the platform, share their gains and questions with each other, and then bring the questions they encountered before class to the classroom for an exploratory attempt.

Teachers will teach the content produced into micro-video for students to watch in advance, which allows physical education teachers to spend more time on classroom teaching in the organization of classroom activities, and effectively help students to solve the problems that have not been solved in watching the video.

In the practice of student knowledge, teachers can focus on the micro-video of some of the technical movements of the focus and difficulties, random sampling of students for the action of the show, fully understand the students’ understanding of the mastery of the technical movements, so that students enter the classroom with the problem of targeted knowledge testing, through the form of students’ independent thinking to play a guiding, assisting the effect of the form.

In the aspect of independent inquiry, students around the learning tasks set by the teacher, take the initiative to explore the learning and practice, in the process of completing the teaching tasks set by the teacher to master the basic technical movements.

In order to ensure the effective implementation of group activities, teachers must group students in advance. In this study, students were divided into 12 groups, each consisting of nearly 4 people. Teachers according to the problems of different groups to personalized guidance to the group, guiding them to actively seek solutions, tailored to the needs of the students.

In terms of group presentation and communication, group presentation and communication are very important teaching methods in the flipped classroom as a method of checking the results of physical education course learning. He can not only flip the role change between teachers and students, stimulate students’ learning motivation and the positive degree of bold expression, but also stimulate healthy competition between students, so that students can fully display their learning achievements, realize the collision of thinking, and enhance the learning effect of the classroom.

Teachers correct and improve the previously recorded teaching micro-video through the students’ completion of specific movement skills in teaching practice, focusing on the problems encountered by the students in the practice of technical movements and the frequency of erroneous movements, and correcting the video information to provide experience for the implementation of the teaching process in the future.

Students can analyze their own mastery of movements against their own actual situation through the teacher’s corrected teaching video, which provides a good way to improve their technical movements. After the lesson, teachers and students can use the communication platform to exchange and communicate about the situation during the lesson. After communicating with the teachers, the students will practice and revise purposefully against the completion of their own technical movements and their own specific problems, so as to find problems and solve problems in time, and to provide a basis for the internalization and enhancement of their own technical movements.

In the new era, college students are rich in learning tasks and social activities, and the work and rest situation varies greatly among different students. They are energetic and full of curiosity about new things, and novel fitness programs can stimulate the enthusiasm of college students for exercise. It is conducive to maintaining the effectiveness and continuity of fitness. On the basis of continuous teaching reform, the teaching methods of college physical education courses tend to be diversified, and students are more flexible in choosing courses. The learning paths for sports skills in different courses are obviously different. Therefore, when formulating personalized exercise prescription for college students, not only should it take into account the learning of sports option skills and the promotion of physical fitness and health, but it must also have the characteristics of high efficiency, precision, diversity, and innovation, so as to accurately buttress the goals of college physical education teaching and students’ personalized physical exercise needs.

This paper proposes to construct a personalized exercise prescription system for college students based on constructivist learning theory and using modern information technology such as artificial intelligence. Constructivist learning theory believes that learning is to guide the learner to grow (construct) new experiences from the original experience. The development of personalized exercise prescriptions for college students is also a process of knowledge construction, where students’ physical condition and fitness needs are analyzed. Construct the attributes of various resources serving fitness, construct the association relationship between students and fitness resources based on prescription rules to form personalized exercise prescription, and then construct new rules based on the actual effect of exercise prescription. The exercise prescription system must create the exercise prescription user model and resource model. 1)

Exercise prescription user model

The user model of exercise prescription is used to characterize the sports quality of college students, which can be obtained through user modeling. Display. The user model can obtain, display, store, and modify the physical fitness and sports skill levels of students. When constructing the sports prescription user model, different stages of multi-channel, multi-mode fusion approach is used, and the final characteristics of the information obtained include the user’s basic information, physical condition, sports conditions, sports goals, sports preferences, sports ability and so on.

The Mahalanobis distance was proposed by the statistician Mahalanobis, it is a method that can effectively calculate the similarity of two unknown sample sets, unlike the Euclidean distance, it takes into account the connection between the features and the features are scale-independent, for example, the vector composed of body weight and height is treated the same in the Euclidean distance, which is not appropriate. In contrast, the Mahalanobis distance is not affected by the scale, the Mahalanobis distance between two points is not related to the unit of measurement of the original data, and its calculation is based on the overall sample. The formula for calculating the Mahalanobis distance is shown in equation (1). (1)DM(x,y)=(x−y)T∑−1(x−y)

where Σ is the covariance matrix of the multidimensional variables and must be full rank. In terms of student characterization, the conventional measures that can be used for student similarity calculation are weighted Euclidean distance and Mars distance, but it is more difficult to assign the weights due to the large number of student features and the weights of their features can be different under different sports goals. The computation of the Mahalanobis distance is based on the overall sample, eliminating the problem of magnitude, but the computation of the Mahalanobis distance is unstable due to the instability of the covariance.

Considering that the purpose of sportsmen similarity calculation is to find sportsmen whose sports effect and satisfaction are similar under the same sports conditions, sports effect and satisfaction can be utilized as the evaluation index of sportsmen similarity, so this paper adopts artificial neural network to judge the similarity of two sportsmen. Where the model input is S(S1,S2,…,Si,…,Sn) , S_i represents the Euclidean distance between two sportsmen x1(x11,x12,…,x1i,…,x1n) and x2(x21,x22,…,x2i,…,x2n) on feature i, which is calculated as shown in equation (2), and the output P represents the degree of deviation of the total score of two sportsmen’s cases G. (2)Si=(x1i−x2i)2 (3)P=(G1−G2)2G1 (4)G=αE+βMα+β

Eq. (4) where α is the effect preference of the case campaigner, β is the satisfaction preference of the case campaigner, E is the effect of the case, and M is the satisfaction of the case. The schematic diagram of similarity model construction is shown in Figure 2.

Figure 2.

Schematic diagram of similarity model construction

Among them, the sportsmen feature vectors include height, weight, cardiorespiratory capacity, intensity preference, time preference, frequency preference. In the process of model training, the selection of sportsmen training data should be in accordance with the screening of the case data, that is, to ensure that the sports goals of the two sportsmen cases are the same and the sports parameters are similar, and the sports parameters mainly include: sports intensity I, sports time T and sports frequency F, and the similarity of the sports parameter calculations and discrimination conditions are shown in Equation (5). (5){ I=(I1−I2)2,I<=0.02 T=(T1−T2)2,T<=3 F=(F1−F2)2,F<=1

The selection of exercise prescription case data as the basis for the determination of the core parameters of the prescription is inevitably the quality exercise prescription cases generated by similar exercisers. The process of screening quality cases in this paper is shown in Figure 3, respectively:

Figure 3.

Schematic diagram of high quality case screening process

Step 1: Calculate the similarity between the current sportsperson and other sportspersons in the sportsperson information base by means of a similarity model to obtain a set U₁ of sportspersons, and then sort the sportspersons in set U₁ according to the similarity.

Step 2: Obtain the prescription cases corresponding to the athletes in set U₁ in turn, and filter the cases, with two main filtering criteria: first, the goals and types of the prescription cases should be the same as those of the current athlete; second, the overall completion degree of the prescription cases should be more than 80%. Finally, the first k eligible cases are sequentially stored in the similar case set R and the student set U₁ is updated.

Step 3: Filter the similar case collection R based on the exercise effect and exercise satisfaction to obtain its quality case collection R^*, whose filtering conditions are as in equation (6), where E indicates the effect of the filtered exercise case, 0.7e indicates the 70% effect in R, M indicates the satisfaction of the filtered exercise case, and 0.7m indicates the 70% satisfaction in R. (6){ E>=0.7e,e∈R M>=0.7m,m∈R

In this paper, based on the sports prescription case data and sportsmen’s preference characteristic parameters, the core parameter space of sports prescription is used as a constraint to recommend more targeted sports prescription parameters with higher prescription effect and satisfaction for users. In traditional recommendation ideas, users with similar characteristics are generally recommended based on the degree of case access. If this recommendation approach is borrowed, it is necessary to find cases that are similar to the current exerciser and have higher effect and satisfaction of the resulting exercise prescription cases, and recommend the obtained exercise prescription case parameters to the current exerciser. Therefore, this paper designs a parameter recommendation method of fusion calculation of core parameters of exercise prescription, which fuses all the high-quality case parameters that satisfy the conditions, so as to improve the effect of prescription parameters and ensure the stability of prescription parameters at the same time. Its overall process is shown in Figure 4.

Figure 4.

Schematic diagram of the core parameter fusion process

Where F represents the space of frequency parameter constraints for exercise prescription, where the overall effect discrimination of the preselected frequency z is based on similar case versus quality case data, calculated as shown in Equation (8). (8)f(z)=∑j=1kiSim(u,uj)×(αej+βmj)h*(z−f)k+1×ln(1+h*n*) ej∈R*,mj∈R*

Where Sim(u,uj) is the degree of similarity between the current user and the user of case j, with higher values representing greater similarity, α is the effect preference of the current campaigner, e_j is the effect level achieved by frequency z in the j th case of the quality case, β is the satisfaction preference of the current campaigner, m_j is the satisfaction level achieved by frequency z in the j th case of the quality case, h^* is the number of frequency z in the quality case R^*, f is the user’s expected frequency, k is the user’s frequency preference factor, n^* is the number of frequencies z in similar cases R, h*/n* is the frequency z in the case quality rate, and ln(1+h*/n*) serves to penalize frequencies with low case quality rates. Finally, the exercise frequency z with better comprehensive effect is selected by discriminative calculation, and the case with exercise frequency z in the quality case collection R^* is deposited in the case collection R1* . 2)

Exercise prescription intensity clustering: after dividing the high-quality case set by pre-selected frequencies, the exercise intensity in case R1* is clustered and analyzed by using clustering algorithm.

Where q_j represents the exercise volume of a single exercise prescription case in case set R2* , calculated from the exercise intensity and exercise time in the exercise prescription case. Its exercise intensity I is calculated as shown in equation (11): (11)I=∑j∈Ri*(αej+βmj)2×ij∑j∈Rz*(αej+βmj)2,ej∈R2*,mj∈R2*

Where i_j represents the exercise intensity of a single prescription case in case set R2* , described by the percentage of maximum heart rate of the exerciser. Finally, the exercise time was determined in based on the relationship between exercise volume, exercise intensity, and exercise time in the sports domain, i.e., exercise time = exercise volume/exercise intensity.

In order to investigate the reasons for the non-significant differences in the experimental subjects’ attitudes toward physical activity, homogeneity tests were conducted on each of their eight sub-dimensions, and the results are shown in Table 1:

Table 1 shows that the p-value is greater than 0.05 for all eight sub-dimensions and therefore all sub-dimensions are not significantly different.

After a 4-week teaching experiment, the changes in the physical activity level of the subjects were tested. In this study, the paired samples t-test and independent samples t-test were conducted respectively, so as to analyze the role of the two teaching modes on the influence of this index and whether they have a significant effect or not, and the results are shown in Table 2:

Table 2 “Behavioral Attitude” shows that the average level before the experiment is 28.283, and the average level after the experiment is 30.545, which is tested to be P>0.05, indicating that the level of “Behavioral Attitude” of the students in the experimental class after the teaching experiment has been improved but not significantly different. It means that the level of “behavioral attitude” of the students in the experimental class has been improved after the teaching experiment, but there is no significant difference. The flipped classroom teaching model attaches importance to cultivating students’ awareness of lifelong physical activity, but due to limited experimental conditions and time, it cannot achieve significant results. Therefore, the effect of the flipped classroom teaching model on “behavioral attitudes” is relatively small.

Table 2 “Target Attitude” shows that the average level before the experiment was 45.342 and 49.254 after the experiment, and the P<0.05 was obtained, indicating that the level of “target attitude” of the students in the experimental class was significantly improved after the teaching experiment, and there was a significant difference. The flipped classroom teaching model is designed to stimulate students’ enthusiasm for learning and increase their motivation for what they have learned, so it has a significant effect on improving the level of “target attitude”.

Table 2 “Behavioral Cognition” shows that the average level before the experiment was 25.343 and 28.542 after the experiment, and the P<0.05 was obtained, indicating that the level of “behavioral cognition” before and after the experiment was greatly improved, and there was a significant difference. In summary, the flipped teaching mode is conducive to the improvement of the level of “behavioral cognition” in “physical exercise attitude”.

Table 2 “Behavioral Habits” showed that the average level before the experiment was 36.967 and 40.254 after the experiment, and the P<0.05 was obtained. Therefore, the flipped classroom teaching mode is conducive to the improvement of students’ “behavior habits”.

Table 2 “Behavior Intention” shows that the average level before the experiment was 28.452 and 32.813 after the experiment, and the P<0.05 was obtained, indicating that the level of “behavior intention” of the students in the experimental class was significantly improved after the teaching experiment, and there was a significant difference before and after the experiment. Because the flipped classroom teaching mode pays more attention to the diversity of students and the diversity of teaching, after the teacher’s teaching assistance, it can be targeted to different students to implement different teaching methods, students choose the training content according to their own situation, and focus entirely on how to perceive the movement and the changes in themselves, and connect knowledge with life, and combine learning in class with practice after class, which improves students’ learning interest, expanding students’ practice time and promoting students’ favorability to physical exercise. In conclusion, the flipped classroom teaching mode is conducive to the improvement of students’ “behavioral intention”.

Table 2 “Emotional Experience” shows that the average level before the experiment was 40.523 and 43.275 after the experiment, and the P<0.015 was obtained, indicating that the level of “emotional experience” before and after the experiment was significantly improved and there was a significant difference. The flipped classroom teaching mode increased the students’ independent practice time and teacher-student interaction time, which effectively improved their inner feelings level. Therefore, the flipped classroom teaching mode is conducive to the improvement of students’ “emotional experience”.

Table 2 “sense of behavioral control” shows that the average level before the experiment is 23.953, and the average level after the experiment is 29.073, which is tested to be P<0.05 and P<0.01, proving that the “sense of behavioral control” level of the students in the experimental class has been very significantly improved after the teaching experiment. The level of behavioral control among the students in the experimental class has significantly improved, and there is a significant difference. Flipped classroom teaching mode can break through the limitations of the teaching environment, so that students can open the learning activities at any time regardless of the limitations of time and space, with a high degree of autonomy and flexibility. In conclusion, the flipped classroom teaching mode is conducive to the improvement of the level of “behavioral control”.

Table 2 “Subjective Standard” shows that the average level before the experiment was 20.568 and 22.644 after the experiment, and the P>0.05 was obtained, indicating that the “subjective standard” level of the students in the experimental class was not significantly improved after the teaching experiment, and there was no significant difference. The flipped classroom teaching model has little and no obvious improvement in the level of students’ “subjective standards”.

In summary, the flipped classroom teaching mode can effectively improve the level of students’ “physical exercise attitude” and its eight sub-dimensions, among which there are significant differences in “goal attitude”, “behavioral cognition”, “behavioral habits”, “behavioral intention”, “emotional experience” and “behavioral control feeling”, while there is no significant difference between “behavioral attitude” and “subjective standard”.

First, the results of the experiment were analyzed by analyzing the results of the experimenter users in both cases of ontological reasoning and without ontological reasoning. The results of the experiment are shown in Fig. 5, where K is the first K motors in the set of similar motors that were selected.

Figure 5.

Comparison experiment results with ontological reasoning

The experimental results show that with the same number of K-nearest neighbors, the metric of using ontological reasoning followed by similarity calculation performs better in terms of accuracy. The use of ontological similarity metrics can better capture the implicit relationships and features between the campaigners, which improves the accuracy and reliability of the similarity metrics. Such results further validate the effectiveness of ontological reasoning in recommender systems and provide more reliable technical support for exercise prescription recommendations.

In order to verify the effectiveness of the ontological reasoning exercise prescription recommendation model, the following two comparison algorithms are selected in this chapter:

ResNet-EP: This method utilizes one-dimensional residual neural network technique to recommend exercise prescription based on fitness test data.

Collaborative filtering recommendation method: it is a common recommendation system technique that utilizes historical interaction information between users and items to make recommendations.

In this chapter, 10 students are randomly selected in the test set, and each student is tested under the same experimental environment, and the ontology-based reasoning recommendation method, ResNet-EP method and collaborative filtering recommendation method are utilized to recommend exercise prescriptions for these 10 students, respectively. The experimental results are shown in Fig. 6, which demonstrates the exercise effect ratings of the 10 students after implementing the exercise prescription under different recommendation algorithms.

Figure 6.

Comparison experiment results of exercise effects

From the experimental results, it can be seen that the model proposed in this paper obtains relatively high ratings on most users, with an average value of 4.04, and the other two algorithms have an average value of 3.75 and 3.10, respectively.This indicates that the model in this chapter has a higher credibility and effectiveness in recommending a better exercise effect on the majority of users. It can be concluded that the ontology-based reasoning exercise prescription recommendation model has high accuracy and effectiveness in recommending exercise effects, and has better performance performance compared to other algorithms, while it may be more suitable for the exercise recommendation task in practical applications, and can provide users with a better exercise experience.