1 Introduction

With the continuous development of modern information technology, numerous new teaching methods have emerged, including the Flipped Classroom (FC) method [1]. It can enhance the interaction between teachers and students and the students' desire to learn Japanese, thereby enhancing the effectiveness of Japanese instruction. This paper aims to provide theoretical support for Japanese teaching reform and enhance the efficacy of Japanese instruction by analyzing the feasibility of the application of FC in Japanese teaching and examining the application strategies of FC in Japanese instruction. Under the FC teaching model, students can control the video's playing time and progression based on their individual learning circumstances when they watch micro-videos before class. They can skip what they have mastered and repeatedly view what they do not understand. It is important to note that when creating micro-videos, teachers should pay attention to the difficulty of the knowledge, convert simple basic knowledge into videos, and put difficult knowledge in the classroom to be taught, so as to avoid discouraging students' enthusiasm for learning Japanese and to increase the time for interaction between teachers and students after class and students' independent learning, which is beneficial for improving students' learning effect. This paper investigates the evaluation model of the Japanese teaching effect in FC and task-based teaching mode and improves the teaching effect and level of Japanese in FC and task-based teaching mode.

Traditionally, the evaluation methods for assessing the effectiveness of Japanese language teaching in the FC and task-based teaching mode have included approaches such as the multi-source information resource service method, correlation feature extraction method [2], and particle swarm optimization evaluation algorithm [3]. In literature [4], an online and offline hybrid evaluation method for measuring teaching effectiveness is proposed, utilizing big data analysis. However, a challenge arises due to the high computational cost involved in evaluating the Japanese teaching effect within the FC and task-based teaching mode. In literature [5], a model is introduced that recognizes the movement characteristics of Taekwondo to correct students' movements and provides sports guidance and practice through simulation. Nevertheless, this method exhibits limited self-adaptive optimization capability when evaluating the effectiveness of Japanese language teaching under the FC and task-based teaching mode.

To solve the above problems, this paper puts forward an evaluation model of the Japanese teaching effect based on feature offset compensation in flip classrooms and task-based teaching mode. Using distributed mining of association rules to detect the Japanese teaching effect in FC and task-based teaching mode [6], extracting ontology information and association rule items of Japanese teaching effect distribution information in FC and task-based teaching mode, constructing fuzzy decision model of Japanese teaching effect evaluation in FC and task-based teaching mode, and calculating the joint information entropy characteristic value of Japanese teaching effect feature deviation in FC and task-based teaching mode. The feature offset compensation [7] and C-means clustering method [8] are employed to classify and identify the extracted features. By integrating the outcomes of big data mining [9], an intelligent evaluation system for assessing the effectiveness of Japanese language teaching in the FC and task-based teaching mode is achieved. The simulation analysis ultimately demonstrates the excellent performance of this approach in enhancing the accuracy of evaluating the Japanese teaching effect in the FC and task-based teaching mode.

The paper is organized into five chapters to present a coherent and systematic analysis of the research topic. Section 1 is the introduction. Section 2 provides an exhaustive review of the pertinent literature and prior research on the study topic. The proposed methodology is introduced in Sect. 3. The focus of Sect. 4 is the exhaustive testing and evaluation of the proposed method. Section 5 summarizes the study's main findings and draws conclusions based on those findings.

2 Distribution Structure Model and Characteristic Analysis of the Characteristics of Learning Behavior of Japanese Teaching Effect

2.1 Japanese Teaching Effect Characteristic Distribution Model

To realize the evaluation design of the Japanese teaching effect in FC and task-based teaching mode based on big data analysis and the mining of learning behavior characteristics, it is necessary first to build a fuzzy decision model for Japanese teaching effect evaluation in FC and task-based teaching mode, and combine the optimization analysis of distributed structure model to carry out distributed strenuous testing. In the context of big data processing, the characteristics of Japanese teaching effect deviation in the FC and task-based teaching mode include browsing records, web page access records, and multi-source information resource distribution records, among others. The resource information extracted from learning behavior characteristics provides computing, physical, and network ontology resources. Figure 1 depicts the distribution structure description of the Japanese teaching effect in FC and task-based teaching mode.

Fig. 1
figure 1

The distribution structure of Japanese teaching effect in FC and task-based teaching mode

Full size image

The information distribution structure model of Japanese teaching effect in FC and task-based teaching mode, as depicted in Fig. 1, utilizes the multi-source information resource service method for consolidated scheduling and online querying. The MIRaaS platform simplifies multiple sources of resources into a single resource, enabling resource cloud analysis of the Japanese teaching effect in the FC and task-based teaching mode. After parameter estimation, you can use the joint distribution characteristics of the Japanese teaching effect in FC and task-based teaching mode to gather information and construct a fuzzy decision-making model for evaluating the Japanese teaching effect in FC and task-based teaching mode [10]. It is a keyword input for Japanese teaching effectiveness evaluation in FC and task-based teaching mode, then the characteristic quantitative solution of resource distribution is:

$$R_{k} = p(t_{k}^{0} < T_{k} < \lambda t_{k}^{0} ) = F_{Z} \left( {\lambda t_{k}^{0} } \right) - F_{Z} \left( {t_{k}^{0} } \right),$$
(1)

wherein \(t_{k}^{0}\) is the rough set of Japanese teaching effect evaluation, \(T_{k}\) is the time difference of Japanese teaching effect evaluation, \(\lambda\) is the fuzzy degree of Japanese teaching effect evaluation, \(F_{Z}\) is the marginal distribution function of Japanese teaching effect evaluation, and \(\left( {s_{{\text{k}}} ,a_{{\mathbf{k}}} } \right)\) (s1, a1) are the fuzzy closeness vectors between Japanese teaching effect evaluation nodes in the flip classroom and task-based teaching mode [11]. Then, the corresponding binary semantic distribution feature mapping of the Japanese teaching effect in flip classroom and task-based teaching mode is described as follows:

$$\theta :S \to S \times [ - 0.5,0.5]$$
(2)
$$\theta (s_{i} ) = (s_{i} ,0),s_{i} \in S$$
(3)

wherein, \(S\) is the ambiguity of the characteristics of Japanese teaching effect, \(s_{i}\) is the language evaluation parameter of Japanese teaching effect query, and the real number β \(\in\) [0, T] is set as the similarity. The language evaluation set S of Japanese teaching effect query in FC and task-based teaching mode represents the information set of Japanese teaching effect in FC and task-based teaching mode. The \(\{ v_{1} ,...,v_{M} \}\) evaluation model expresses the satisfaction of the results of the Japanese teaching effect evaluation in the FC and task-based teaching mode, and uses the correlation fusion method to carry out big data fusion scheduling [12, 13]. The adaptive information forwarding mapping for Japanese teaching effect evaluation in the FC and task-based teaching mode is described as follows:

$$C_{\Phi } \left( {\mathbf{u}} \right) = \Phi (\Phi^{ - 1} \left( {u_{1} } \right),......\Phi^{ - 1} \left( {u_{n} } \right)), {\mathbf{u}} \in I^{n}$$
(4)

wherein ui is an arbitrary evaluation index weight, φ() is the normal distribution function of learning behavior in Japanese teaching effect in FC and task-based teaching mode, and φ − 1 is the inverse function of φ(). By calculating the comprehensive relative closeness of Japanese teaching effect evaluation in FC and task-based teaching mode, the correlation and fusion characteristics of learning behavior mining in Japanese teaching effect in FC and task-based teaching mode of φ() are obtained.

$$\phi (t) = \left( {2\pi } \right)^{{ - \frac{n}{2}}} \left| \sum \right|^{{ - \frac{1}{2}}} \exp \left[ { - \frac{1}{2}\left( {{\mathbf{t}} - {{\varvec{\upmu}}}} \right)^{^{\prime}} \sum^{ - 1} \left( {{\mathbf{t}} - {{\varvec{\upmu}}}} \right)} \right],$$
(5)

wherein \({\mathbf{t}}\) is the quantitative time parameter of resource distribution, and \({{\varvec{\upmu}}}\) is the fuzzy edge distribution coefficient. The distributed mining of association rules is used to test the teaching effect of Japanese in FC and task-based teaching mode.

2.2 Flip Classroom and Task-Based Teaching Mode of Japanese Teaching Effect Analysis

Construct the distribution structure model of the Japanese teaching effect in the FC and task-based teaching mode, extract the ontology information and association rules of Japanese teaching effect distribution information in the FC and task-based teaching mode, and get the feature set of association rules as follows:

$$\begin{gathered} F_{Z} \left( {\lambda t_{k}^{0} } \right) = p\left( {T_{1} + T_{2} + \cdot \cdot \cdot T_{n} \le \lambda t_{k}^{0} } \right) \hfill \\ {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} = \mathop {\iint { \cdot \cdot \cdot }}\limits_{{Z \le \lambda t_{k}^{0} }} \int {f\left( {t_{1} {,}{\kern 1pt} {\kern 1pt} t_{2} {\kern 1pt} \cdot \cdot \cdot t_{n} } \right)} {\text{d}}t_{1} {\text{d}}t_{2} \cdot \cdot \cdot {\text{d}}t_{n} \hfill \\ \end{gathered}$$
(6)
$$\begin{gathered} F_{Z} \left( {t_{k}^{0} } \right) = p\left( {T_{1} + T_{2} + \cdot \cdot \cdot T_{n} \le t_{k}^{0} } \right) \hfill \\ {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} = \mathop {\iint { \cdot \cdot \cdot }}\limits_{{Z \le t_{k}^{0} }} \int {f\left( {t_{1} {,}{\kern 1pt} {\kern 1pt} t_{2} {\kern 1pt} \cdot \cdot \cdot t_{n} } \right)} {\text{d}}t_{1} {\text{d}}t_{2} \cdot \cdot \cdot {\text{d}}t_{n} \hfill \\ \end{gathered}$$
(7)

wherein Z = t1 + t2 + tn indicates the correlation degree of evaluation nodes of Japanese teaching effect in FC and task-based teaching mode. The detection function is:

$$f(x) = {\text{SINR}} - 20\exp \left( { - 0.2\sqrt {\frac{1}{D}\sum\limits_{d = 1}^{D} {x_{{\text{d}}}^{2} } } } \right) - \exp \left( {\frac{1}{D}\sum\limits_{d = 1}^{D} {\cos (2\pi x_{{\text{d}}} )} } \right)$$
(8)

wherein, SINR is the detection threshold of Japanese teaching effect learning, \(\beta \ge 1\), which represents the deviation coefficient of Japanese teaching effect learning evaluation \(\alpha \in (2,6)\), it is used Gaussian Copula function to calculate fuzzy closeness [14, 15], and get the distribution rules of learning behavior characteristics:

$$\begin{gathered} S_{msg}^{\left( i \right)} \left( {t + \Delta t} \right) = f\left( {S_{msg}^{\left( A \right)} \left( t \right),S_{msg}^{\left( 1 \right)} \left( t \right),...,S_{msg}^{\left( n \right)} \left( t \right),S_{ctxt}^{\left( 1 \right)} \left( t \right),...,S_{ctxt}^{\left( n \right)} \left( t \right)} \right),\forall i \in \left\{ {A,1,...,n} \right\} \hfill \\ S_{ctxt}^{\left( i \right)} \left( {t + \Delta t} \right) = f\left( {S_{ctxt}^{\left( A \right)} \left( t \right),S_{ctxt}^{\left( 1 \right)} \left( t \right),...,S_{ctxt}^{\left( n \right)} \left( t \right)} \right),\forall i \in \left\{ {A,1,...,n} \right\} \hfill \\ \end{gathered}$$
(9)

wherein, \(S_{msg}^{\left( A \right)} \left( t \right),S_{msg}^{\left( 1 \right)} \left( t \right),...,S_{msg}^{\left( n \right)} \left( t \right),S_{ctxt}^{\left( 1 \right)} \left( t \right)\) is an evaluation fuzzy parameter. The fuzzy scheduling method is used to sample the big data of learning behaviors of Japanese teaching effect in FC and task-based teaching mode, and the limited data set of Japanese teaching effect in distributed structure space in FC and task-based teaching mode is described as follows:

$$\begin{aligned} R_{k} & = p(t_{k}^{0} < T_{k} < \lambda t_{k}^{0} ) = F_{Z} \left( {\lambda t_{k}^{0} } \right) - F_{Z} \left( {t_{k}^{0} } \right) \hfill \\ & = \mathop {\iint { \cdot \cdot \cdot }}\limits_{{Z \le \lambda t_{k}^{0} }} \int {f_{{T_{1} ,T_{2} \ldots T_{n} ,}} (t_{1} ,t_{2} \cdot \cdot \cdot t_{n} )} {\text{d}}t_{1} {\text{d}}t_{2} \cdot \cdot \cdot {\text{d}}t_{n} - \mathop {\iint { \cdot \cdot \cdot }}\limits_{{Z \le t_{k}^{0} }} \int {f_{{T_{1} ,T_{2} \ldots T_{n} ,}} (t_{1} ,t_{2} \cdot \cdot \cdot t_{n} )} {\text{d}}t_{1} {\text{d}}t_{2} \cdot \cdot \cdot {\text{d}}t_{n} \hfill \\ & = \mathop {\iint { \cdot \cdot \cdot }}\limits_{{Z \le \lambda t_{k}^{0} }} \int {\left( {2\pi } \right)^{{ - \frac{n}{2}}} \left| \sum \right|^{{ - \frac{1}{2}}} \exp \left[ { - \frac{1}{2}\left( {{\mathbf{F}}_{{\mathbf{T}}} {\mathbf{(t)}} - {{\varvec{\upmu}}}} \right)^{^{\prime}} \sum^{ - 1} \left( {{\mathbf{F}}_{{\mathbf{T}}} {\mathbf{(t)}} - {{\varvec{\upmu}}}} \right)} \right]\prod\limits_{i = 1}^{n} {f_{{T_{i} }} (t_{i} )} } {\text{d}}t_{1} {\text{d}}t_{2} \cdot \cdot \cdot {\text{d}}t_{n} \hfill \\ & \quad - \mathop {\iint { \cdot \cdot \cdot }}\limits_{{Z \le t_{k}^{0} }} \int {\left( {2\pi } \right)^{{ - \frac{n}{2}}} \left| \sum \right|^{{ - \frac{1}{2}}} \exp \left[ { - \frac{1}{2}\left( {{\mathbf{F}}_{{\mathbf{T}}} {\mathbf{(t)}} - {{\varvec{\upmu}}}} \right)^{^{\prime}} \sum^{ - 1} \left( {{\mathbf{F}}_{{\mathbf{T}}} {\mathbf{(t)}} - {{\varvec{\upmu}}}} \right)} \right]\prod\limits_{i = 1}^{n} {f_{{T_{i} }} (t_{i} )} } {\text{d}}t_{1} {\text{d}}t_{2} \cdot \cdot \cdot {\text{d}}t_{n} \hfill \\ \end{aligned}$$
(10)

wherein, \({\mathbf{F}}_{{\mathbf{T}}} {\mathbf{(t}})\) is the structural distribution factor, and \({{\varvec{\upmu}}}\) is the characteristic quantity of reliability distribution. By adopting the generalized learning algorithm [16], the self-adaptive learning [17] of the personalized evaluation process of Japanese teaching effect in the integrated flip classroom and task-based teaching mode is carried out. C is the cost factor of personalized evaluation of the Japanese teaching effect in the integrated flip classroom and task-based teaching mode. The difference function of personalized evaluation of Japanese teaching effect in the integrated flip classroom and task-based teaching mode is obtained as follows:

$$f(x) = \sum\limits_{i = 1}^{n} {(\alpha_{i} - \alpha_{i}^{ * } )K(x_{i} ,x_{j} )} + b$$
(11)

wherein, \(\alpha_{i}\) and \(\alpha_{i}^{*}\) are individualized attribute values and edge distribution characteristic coefficients that represent the Japanese teaching effect in the integrated flip classroom and task-based teaching mode, \(K(x_{i} ,x_{j} )\) is a symmetric kernel function that satisfies Mercer condition, and \(b\) represents the evaluation threshold. According to the extraction results of the Japanese teaching effect in the FC and task-based teaching mode, the self-adaptive evaluation of the Japanese teaching effect in the FC and task-based teaching mode is carried out [8, 18].

3 Evaluation Model Optimization

3.1 Decision-Making Model of Japanese Teaching Effect Evaluation

This paper optimizes the design of the Japanese teaching effect evaluation model in FC and task-based teaching mode and proposes a Japanese teaching effect evaluation model in FC and task-based teaching mode based on feature offset compensation. The personalized guidance space of the Japanese teaching effect and students' learning behavior in the integrated flip classroom and task-based teaching mode is represented by a nonlinear mapping, and the evaluation information of the Japanese teaching effect in the integrated flip classroom and task-based teaching mode is mapped to a high-dimensional feature space by combining fuzzy decision-making and intelligent swarm optimization [19,20,21]. It is assumed that the training sample set of Japanese teaching effect evaluation under the combination of FC and task-based teaching mode consists of training samples in which the personalized evolution feature quantity serves as the evaluation model's input vector. Combining the preference mining algorithm of learning behavior, the objective function of personalized evaluation of Japanese teaching effect in FC and task-based teaching mode is:

$$\begin{gathered} \hfill {\textit{minimize}} \;\;\;\;\;\;\;\frac{1}{2}\left\| w \right\|^{2} + C\sum\limits_{i = 1}^{n} {(\xi_{i} + \xi_{i}^{ * } )} \\ \hfill subject\;to\;\;\;y_{i} - (w^{^{\prime}} \Phi (x_{i} ) + b) \le \varepsilon - \xi_{i} \\ \hfill (w^{^{\prime}} \Phi (x_{i} ) + b) - y_{i} \le \varepsilon - \xi_{i}^{*} \\ \hfill \xi_{i} ,\xi_{i}^{*} \ge 0,i = 1,2, \ldots ,n; \; C > 0 \\ \end{gathered}$$
(12)

In which, \(\xi_{i}\) and \(\xi_{i}^{*}\) represent the dimensions and convolution of information subspace, and a hybrid kernel function of Japanese teaching effect evaluation under the FC and task-based teaching mode is constructed. Its expression is:

$$K_{\min } = \beta K_{poly} + (1 - \beta )K_{rbf} ,{\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} \beta \in (0,1)$$
(13)

wherein, \(K_{poly} = [(x \cdot x_{i} ) + 1]^{2}\) represents the kernel function of personalized preference of Japanese teaching effect, extracts the ontology information and association rule items of Japanese teaching effect distribution information in FC and task-based teaching mode, and constructs the fuzzy decision model of Japanese teaching effect evaluation in FC and task-based teaching mode, and obtains the fuzzy decision function as follows:

$$\begin{aligned} f_{\lg - M} (z) & = (f_{\lg } (z),f_{\lg - x} (z),f_{\lg - y} (z)) \\ & = (f_{\lg } (z),h_{x} * f_{\lg } (z),h_{y} * f_{\lg } (z)) \\ \end{aligned}$$
(14)

In the above formula, \(f_{\lg } (z)\) represents the scoring value of learning behavior items in the evaluation of Japanese teaching effect under the FC and task-based teaching mode.

3.2 FC and Task-Based Teaching Mode to Optimize the Output of Japanese Teaching Effect Evaluation

Based on constructing the fuzzy decision-making model of Japanese teaching effect evaluation in FC and task-based teaching mode, the joint information entropy eigenvalue of Japanese teaching effect feature deviation in FC and task-based teaching mode is calculated, and the time series of feature distribution is \(x(t)\), \(t = 0,1, \ldots ,n - 1\). Under the constraint rules of association rules, the regional distribution function of personalized preference information structure of Japanese teaching effect in FC and task-based teaching mode is obtained as follows:

$$\begin{aligned} E^{cv} (c_{1} ,c_{2} ) & = \mu \cdot {\text{Length}}(C) + \nu \cdot {\text{Area}}({\text{inside}}(C)) \\ & \quad + \lambda_{1} \int_{{{\text{inside}}(C)}} {\left| {I - c_{1} } \right|}^{2} {\text{d}}x{\text{d}}y + \lambda_{2} \int_{{{\text{outside}}(C)}} {\left| {I - c_{2} } \right|}^{2} {\text{d}}x{\text{d}}y \\ \end{aligned}$$
(15)

wherein, \(c_{1}\) and \(c_{2}\), respectively, represent the adaptive feature matching feature coefficients for the evaluation of personalized preference information of Japanese teaching effect under the integrated flip classroom and task-based teaching mode, \({\text{Length}}(C)\) represents the length coefficient of personalized preference information of Japanese teaching effect under the integrated flip classroom and task-based teaching mode to be evaluated, and \(Area(inside(C))\) represents the regional feature distribution. Let the membership degree of the first class of Japanese teaching effect data stream in the identification concept in the current FC and task-based teaching mode be recorded as \(CF = \left\langle {F,Q,n,RT_{1} ,RT_{2} ,RW} \right\rangle\), where the metadata compatible mapping relation is \(\left\{ {X_{1} ,X_{2} ,..,X_{n} } \right\}\) as the correlation constraint coefficient, and calculate the joint information entropy characteristics of Japanese teaching effect distribution information in FC and task-based teaching mode, which is expressed as:

$$\begin{aligned} J_{I} \left( {nT_{{\text{B}}} } \right) & = \frac{2\sqrt J }{{SN}}\sin c\left( {\pi \Delta fT_{{\text{C}}} } \right) \\ & \quad \times \sum\limits_{i = 0}^{N - 1} {c_{i} \cos \left[ {2\pi \Delta fT_{{\text{C}}} \left( {nN + i + \frac{1}{2}} \right) + \varphi_{j} } \right]} \\ \end{aligned}$$
(16)

wherein, \(J\) is the process model coefficient for designing and executing the FC teaching mode, \(T_{{\text{C}}}\) is the reliability parameter for students to become innovative technical talents, and \(c_{i}\) is the characteristic value of behavior integration. The joint distribution density function of learning behavior characteristics under FC and task-based teaching mode is described as:

$$\begin{gathered} f_{{T_{1} ,T_{2} , \ldots T_{n} ,}} (t_{1} ,t_{2} \ldots t_{n} ) \hfill \\ = c\left( {F_{{T_{1} }} (t_{1} ),F_{{T_{2} }} (t_{2} ), \ldots F_{{T_{n} }} (t_{n} )} \right)\prod\limits_{i = 1}^{n} {f_{{T_{i} }} (t_{i} )} \hfill \\ {\kern 1pt} { = }\left( {2\pi } \right)^{{ - \frac{n}{2}}} \left| \sum \right|^{{ - \frac{1}{2}}} \exp \left[ { - \frac{1}{2}\left( {{\mathbf{F}}_{{\mathbf{T}}} {\mathbf{(t)}} - {{\varvec{\upmu}}}} \right)^{^{\prime}} \sum^{ - 1} \left( {{\mathbf{F}}_{{\mathbf{T}}} {\mathbf{(t)}} - {{\varvec{\upmu}}}} \right)} \right]\prod\limits_{i = 1}^{n} {f_{{T_{i} }} (t_{i} )} \hfill \\ \end{gathered}$$
(17)

wherein, \({\mathbf{F}}_{{\mathbf{T}}} \left( {\mathbf{t}} \right) = (F_{1} \left( t \right), \, F_{2} (t)...F_{N} \left( t \right))\), at this time, the compactness of metadata corresponding to the characteristic deviation features of Japanese teaching effect in FC and task-based teaching mode is reversed, a fuzzy decision model for Japanese teaching effect evaluation in FC and task-based teaching mode is constructed, the joint information entropy characteristic values of Japanese teaching effect characteristic deviation in FC and task-based teaching mode are calculated, and the extracted characteristic quantities are classified and identified by using characteristic deviation compensation and C-means clustering method [22, 23], so as to realize the evaluation of Japanese teaching effect in FC and task-based teaching mode. The flowchart of the realization is shown in Fig. 2.

Fig. 2
figure 2

Implementation process of improved algorithm

Full size image

4 Simulation Experiment and Result Analysis

A simulation experiment was conducted to test the performance of this method's application in realizing the evaluation of the Japanese teaching effect under the FC and task-based teaching mode. Using the example of a basic Japanese course, this study selected students from a practical Japanese class and divided them into experimental and control groups. There are 41 classes in total, with 8 boys and 33 girls. There are 37 control classes, with five boys and 32 girls. The two classes share the same class time, class duration, and instructional materials. The experimental class adopts the flip classroom teaching mode for design and implementation throughout the semester, whereas the control class adheres to the traditional teaching mode, utilizing the blackboard and multimedia courseware to aid instruction. The final unified examination paper is adopted for the examination without reference materials.

Matlab 7 is used to design the experiment, and the cloud-based Pearson Database provides the data for the Japanese teaching effect evaluation system in FC and task-based teaching mode. The size of the sample set is 2000, while the size of the training data set is 120. The sample length of Japanese teaching effect characteristic distribution data in FC and task-based teaching mode is 400, the personalized evolution characteristic dimension of Japanese teaching effect distribution in FC and task-based teaching mode is 4, the sampling time delay is 0.45, the convergence judgment threshold is 0.28, and the measurement value of Japanese teaching effect character is 1. For autonomous learning, students in the experimental class utilize common electronic platforms such as Weixin, QQ Group, QQ Shared Files, etc. to download pre-recorded teaching videos and teachers' prepared learning plans prior to class. Typically, video length ranges between 10 and 20 min. During this time, students' concentration is at its peak. Students with access to computers, tablets, and smartphones must watch and study in advance and learn independently, step-by-step, per the study plan. Students who lack these other functional devices can watch and study using multimedia devices provided by schools, such as libraries and computer labs. Teachers conduct formative assessments on students and give feedback to answer questions through online testing and other software during the teaching process. Finally, at the end of the semester, they conduct final assessments on students and combine various assessment methods to evaluate students' learning effects in the experimental class.

Figure 3 demonstrates that the original data of learning behaviors are not regular, and the feature points are dispersed; therefore, it is necessary to cluster and combine the data. The method in this paper is used to extract the features after mining learning behaviors, and the joint information entropy features of Japanese teaching effect distribution information in the FC and task-based teaching mode are extracted. Figure 4 displays the results of the feature clustering analysis.

Fig. 3
figure 3

Data mining results of learning behavior

Full size image
Fig. 4
figure 4

Cluster analysis results of learning behavior characteristics

Full size image

From the analysis of Fig. 4, it can be seen that the method of this paper can effectively reflect the preference information of learning behavior in the Japanese teaching effect under the mode of merging and flipping the classroom and task-based teaching, and test the convergence curve, as shown in Fig. 5.

Fig. 5
figure 5

Convergence curve test

Full size image

According to the analysis of Fig. 5, this method can effectively realize the evaluation of the Japanese teaching effect, and it has good convergence. With this method, we can guide the assessment of the Japanese teaching effect under the FC and task-based teaching mode, and the comparison results of the accuracy of the evaluation are shown in Fig. 6.

Fig. 6
figure 6

Comparison of evaluation accuracy

Full size image

From the analysis of Fig. 6, it can be seen that the accuracy of Japanese teaching effect evaluation in FC and task-based teaching mode is high, and the confidence level is good. The statistical average of evaluation satisfaction level is 0.932, which is 14.5% higher than the traditional method, which shows the superiority of this method. Test the time cost of Japanese teaching effect evaluation in flip classroom and task-based teaching mode by different methods, and get the comparison results in Table 1. The analysis shows that this method's real-time performance of Japanese teaching effect evaluation in flip classroom and task-based teaching mode is better, and the time cost is shorter.

Table 1 Time cost comparison (unit: ms)
Full size table

5 Conclusions

In conclusion, the FC is a student-centered approach where students learn and communicate independently. However, this does not eliminate the role of teachers; their role shifts from classroom leaders to organizers and coordinators, guiding students' independent learning. Teachers must regulate the classroom rhythm and appropriately group students based on their levels and abilities. Grouping allows for mutual learning among students. Group leaders are supervisory, relieving teachers from individually monitoring each student. Due to the nature of Japanese learning, where students typically start with minimal knowledge, essential components such as listening, speaking, reading, and memorization are indispensable. However, if teachers individually supervise and assess each student, time allocation becomes an issue. In such cases, group leaders can take over the responsibility of checking recitations and ensuring that group members follow the learning plan set by the teacher. They also guide their team members in dividing tasks and adopting effective learning methods. When group discussions become intense or difficult to sustain, teachers should timely intervene and encourage students to maintain their enthusiasm for learning.

It is crucial to develop an effective evaluation model for assessing the Japanese teaching effect in the FC and task-based teaching mode. This will improve the quality of Japanese teaching and enhance both teachers' satisfaction with their teaching and students' learning behavior. In this paper, we proposed an evaluation model based on feature offset compensation for assessing the Japanese teaching effect in this mode. We employed distributed mining of association rules to detect the teaching effect and extracted ontology information and association rules related to the distribution of the Japanese teaching effect. Additionally, we constructed a fuzzy decision-making model for evaluating the teaching effect. We classified and identified the extracted features using feature deviation compensation and the C-means clustering method by calculating the joint information entropy characteristic value of the teaching effect deviation. This approach allowed us to accurately evaluate the Japanese teaching effect in the FC and task-based teaching mode. Our research demonstrates that this method improves the accuracy of mining and clustering the writing characteristics of learning behaviors, resulting in high satisfaction and confidence levels in the evaluation process.

Refining and validating the proposed evaluation model in future work is essential. Additionally, investigating additional factors that may influence the Japanese teaching effect in the FC and task-based teaching mode would enhance our understanding of the teaching and learning dynamics. Conducting empirical studies and gathering feedback from teachers and students would provide valuable insights for improving the evaluation model and refining the instructional strategies for Japanese language education in this innovative teaching approach.