Skip to main content

Education on quality assurance and assessment in teaching quality of high school instructors

Abstract

Research on language teaching quality has certainly stood out enough to be noticed as the momentum higher school teaching change proceeds to extend and develop. The way to further developing language teaching quality is to further develop teaching quality, and educator assessment is a significant instrument for doing as such. Accordingly, school administration necessitates the turn of events and refinement of a framework for assessing language teaching quality. Thus, hybrid learning technique for assessing the teaching quality of high school instructors should be created. We present an interesting model for assessing the quality of homeroom teaching involving artificial intelligence innovation in high schools, which depends on better hereditary calculations and neural networks. The fundamental thought is to utilize higher request factual elements (skewness, change, second and kurtosis), even vulnerability, Improved Independent Component Analysis (IICA), Holo-entropy based highlights to remove the underlying loads and limits of gathered data. The teaching quality assessment results were enhanced by further developing the neural network’s forecast accuracy and intermingling speed, bringing about a more down to earth plot for assessing high school language teaching quality. We have led simulation investigations and comparative analysis utilizing the Bi-directional Long Short Term Memory (Bi-LSTM) and Convolutional Neural Network (ConvNet/CNN) models. Then, an education quality assessment framework is laid out by hybrid optimizing model parameters which is Seagull Optimization Algorithm (SOA) and Red Colobuses Monkey (RCM).

Introduction

As of now, Education is the certain quality assurance, which is basically because of the advancement of monetary assessment. As a scaffold instrument for correspondence between various nations, language assumes a vital part in worldwide collaboration, data trade, and innovation spread. It is fundamental to evaluate the tutoring quality impartially, figure out the issues in teaching on schedule, and set forward the restorative measures likewise. Thusly, a similar stuff wonderful teaching appraisal framework is a pivotal segment of training work, which can assist the schooling division with evaluating the teaching quality from the parts of language study hall teaching, teaching accomplishments, school the executives, and instructors [1]. To further develop the issues prevailing in education and further develop the teaching quality, it is important to explain the present teaching circumstance. It is important to develop an educational assessment appraisal framework reasonable for the real circumstance of colleges. Oneself situating of colleges regulates that the appraisal of instructors’ teaching quality not exclusively can be surveyed by hypothetical teaching yet in addition can give close consideration to the advancement of understudies’ common-sense abilities so understudies have a strong hypothetical establishment and useful ability that fulfill the interest of social turn of events. Consequently, the evaluation of educational quality is a critical substance of schooling the board [2, 3].

Artificial intelligence innovation is situation off floods of uses in different enterprises [4]. Particularly in the domain of schooling, advances, for example, artificial intelligence instruction stages are turning out to be progressively full grown. Utilizing artificial intelligence innovation to advance schooling and tutoring change to meet the improvement of education modernization has turned into an expansive agreement in the education local area [5,6,7,8]. With the nonstop turn of events and process of network data innovation, the hybrid learning strategy has drawn in increasingly more consideration and consideration. The data age has given an expansive improvement space to hybrid studying [9,10,11,12]. Accordingly, it is of down to earth importance to spread over a hybrid studying method that syndicates a smart studying stage with the customary education of coaching in high schools to change the conventional teaching method. The mixed studying method has been read up abroad for quite a while. So many exploration outcomes have been shaped. This method started from web based studying [1, 12,13,14,15]. As soon as 1996, E-Studying was referenced in the “Preparing Magazine”. Like internet studying, mixed advancing additionally mirrors individuals’ concept of applying data innovation to teaching [2].

In this way, present day data innovation has grown quickly, and a hybrid studying method has been suggested. Mixed studying can be essentially perceived as a combination of two studying modes, one is the customary homeroom studying [3, 16,17,18,19]. Yaqoob accepts that the mixed studying method has an extremely rich implication. The expansive inclusion incorporates four focuses: one is to take on an assortment of schmoozed advances to accomplish instructive objectives; the second is to combine educator direction preparing and present day being taught innovation; the third is utilize current being taught procedures to finish being taught undertakings; Fourth, genuinely take advantage of different being taught strategies and embrace progressed being taught methods to at long last accomplish the best studying impact [4, 20, 21].

Alaei et al., accept that this new studying method is to take full advantage of current education methods to happen the customized studying style, and afterward understand the customized manipulation of education to accomplish the best education impact [5]; Saggi accepts that conventional studying and mixing, there are big contrasts amidst the two methods of studying, the previous necessities to go done five means during the progress to the last option, which are the strengthening mode, the focal market mode, the smorgasbord mode, the replacement method and the totally provocative [6, 22, 23]; Authors accept that hybrid studying the method up close and personal education. It doesn’t need the utilization of pages for education in the homeroom, however it incorporates up close and personal education and internet figuring out how as far as possible, completely the upsides. It is accepted that embracing a hybrid studying method can help correspondence among instructors and understudies [24], [25].

The evaluation of educating manner in colleges is a fluffy and complicated heterogenous inquiry, which is primarily appeared in a wide scope of appraisal substance and an enormous number of markers. Scientists utilize dim social analysis, logical ordered progression process, and fluffy all-round appraisal to evaluate education quality [4,5,6]. Albeit different evaluation meters are completely thought of, it is hard to dispose of partisanship and irregularity, which has specific limits on nonlinear educational quality appraisal. With the progressive extension of PC, for instance, CNN, have showed up in the exploration of tutoring quality appraisal [7]. CNN, which can improve the specific of teaching quality appraisal somewhat. Be that as it may, in light of the fact that the plunge learning strategy took on by neural network is neighbourhood search, it is basic for neural network to separate into nearby least, and its speculation ability is frail. Taking into account these shortcomings, writing [8] chooses the most extreme entropy rule with the capacity of describing vulnerability appropriation data to supplant the blunder standard of CNN algorithm and lays out the greatest appraisal model of educational quality. The overall mistake of the better model is moderately little, and the outcomes strategy has specific speculation. Writing [9] utilizes innovation to lay out a evaluation framework in colleges. The framework accepts educational evaluation rules as info appraisal result. The appraisal result implies might tolerate the flimsy spot of regular assessment quality evaluation implies. Presentation of neural network in instructional quality appraisal as of late, the shortcomings in view of a customary expectation model have been extraordinarily survived, and the outcomes are great [10]. In any case, on the grounds that the customary neural network algorithm is not difficult to be restricted to the base locally, it can’t further develop the combination speed, so it is challenging consequences join precisely.

  • Defines the proposed hybrid classifier is formulated by combining the Bi-LSTM and CNN models is proposed.

  • When relating with other traditional approaches, the replication portraits that the regular appraisal accuracy of the proposed model.

The rest of this paper is arranged as: Sect. 2 discusses the recent works done in IoT. Section 3 discusses about the Proposed Massive IoT energy Optimization Model: An Overview. The recoded outcomes are discussed in Sect. 4. This paper is concluded in Sect. 5.

Evaluation of teaching quality system

The objective of educational quality assessment [1,2,3,4] is to advance schooling change, further develop administration quality, diminish understudy trouble [5,6,7], foster understudies’ intelligence [8], and help understudies assess and take care of issues. We should accomplish the solidarity of philosophy while assessing the quality of teaching, and we should do as such in a goal, fair, and level-headed way, instead of abstractly speculating or blending individual sentiments [9, 10]. In schools and colleges, teaching quality is regularly surveyed on four channels: understudy assessment, master assessment, peer assessment, and educator self-assessment, with the last assessment results integrated. In any case, certain issues stay during the time spent creating, using, and assessing the teaching quality appraisal framework’s assessment results [3, 4], for example, assessment hypothesis research [11], assessment strategy use, assessment technique update, and assessment data analysis. These issues straightforwardly affect educational organizations. Later on, quality appraisal and information extraction will be basic [12]. The markers in the assessment framework by and large include teaching disposition [13], capability in teaching content, and fundamental teaching abilities [14,15,16], and so on In any case, the extensive quality of teachers isn’t just reflected in the above perspectives yet additionally incorporates teachers’ information level, teaching research ability, teaching plan ability, and teachers’ development ability, and so forth In any case, as of now, these assessment lists which can completely mirror the extensive quality and character of teachers are only here and there engaged with the assessment framework, so they ought to be completely thought to be in the foundation of the assessment record framework.

The pointers in the assessment framework [17] of the distinctions in the level of impact of the assessment results should be relegated various loads, yet numerous schools colleges actually utilize a similar weight strategy, or emotionally determined a weight conveyance table to lay out the assessment framework, and afterward utilize this assessment framework to assess. Along these lines, the sensible assignment of weight is the critical stage to consummate the assessment framework. Educational quality assessment frameworks should be corresponding in schools and colleges; the current circumstance requires the development of an educational quality assessment framework; situating determines for teachers teaching quality assessment isn’t possible exclusively through the hypothesis of teaching assessment; more consideration should be paid to the development of understudies. They can likewise address the requirements of social advancement in reality. Accordingly, surveying the executives. We will utilize shrewd innovation to make the assessment of teaching quality more logical and numerical in light of the overhead information. Simultaneously, the review content of this paper fills in as an analysis, input, and impetus, considering the early discovery of challenges in the teaching system and ideal criticism to teachers to endlessly further develop teaching quality. Moreover, logical assessment will apply proper strain to teachers, persuading them to effectively upgrade the quality of their guidance and ability improvement.

Coming up next are the principle advancement: This research fosters a model for assessing high school language teaching quality in view of a better hereditary algorithm and a neural network, bringing about an original way for assessing school teaching quality. Simultaneously, it is relied upon to give a significant reference premise to the teaching the board office to get Artificial Intelligence teaching quality assessment work plans and projects, as well as give sensible decisions to the advancement and assessment of teachers’ expert titles, and make teaching the executives more logical, organized, and standardized. This research utilizes a versatile transformation hereditary procedure to advance the BP neural network’s underlying loads and limits. Since the proposed hybrid classifier is formed by combining the Bi-LSTM and CNN models, using a superior hereditary algorithm to upgrade the underlying loads and edge esteem, diminish the BP neural network’s preparation span to fulfill the edge, and increment the neural network’s teaching quality to the forecast accuracy evaluation discoveries.

Deep learning on teaching quality

With the quick improvement of high school [1,2,3,4], an educational system’s standing has turned into the main rule for understudies while picking a educational system, and an educational system’s standing is generally determined by the quality of its education [5,6,7]. An educational system’s first concern is tutoring quality, since it affects the school’s presence and improvement as well as the future and predetermination of its understudies. In any case, because of the enormous scope enlistment increment of schools and colleges for a long time, various related issues have arisen, including a teacher shortage, a decrease in the quality of understudies, and a shortage of educational and teaching gear and operations offices. These issues have ignited broad worry in the public arena, inciting a conversation turned into a disagreeable issue, yet it has likewise developed into an extensive impression of school and college work. Further developing educational quality is generally an intriguing issue in higher education. “Further developing language teaching in higher education quality” is exceptionally functional. Accordingly, the inside appraisal action puts a solid accentuation on the making of a framework for assessing language teaching quality.

Education is the main part of schoolwork. The adequacy of a school’s education is a critical determinant of its prosperity [10]. Education quality administration is basic to the school’s general quality administration. Assessment of teachers’ education quality guides school pioneers and managers in defining how much education objectives have been met, getting a handle on the school’s education work exhaustively and precisely, and further developing education quality. The subject of education quality assessment [11,12,13] will likewise be examined. This is a basic task. The foundation of education exercises in schools and colleges, as well as the main association in guaranteeing the quality of ability improvement in these organizations, is education quality assessment. Thus, evaluating language schooling assessment is a basic part of advancing high-assessment education, evolving schooling, and further developing schooling assessment. It significantly affects by and large educational assessment. Since schooling is both an otherworldly work and a workmanship, there is no settled recipe. Nonquantitative standards are broadly used in the assessment of schooling assessment, since characterizing what establishes schooling is extremely ambiguous and hard to quantify. The trouble and intricacy of evaluating assessment schooling and learning are both piece of the schooling system. It is undeniably more challenging to assess the assessment of a teacher’s guidance than it is to assess the assessment of an item. The schooling process is comprised of an assortment of parts, and a two-way action includes the two teachers and understudies. There are many elements that impact schooling assessment, and there is something else to research. The improvement of a logical and sensible schooling assessment framework [14] that can precisely and even-handedly evaluate schooling assessment is a huge test.

Since it is a complicated and theoretical nonlinear issue, it is hard to communicate the assessment of schooling assessment at high school utilizing a numerical model or insightful recipe. The neural network model [14,15,16,17] can accomplish shared planning amidst any aspects and has solid nonlinear handling capabilities [18, 19]. Accordingly, making a neural network model to deal with the test of assessing the assessment of school guidance is a viable measurement. PC vision [19,20,21,22] in the assessment of schooling assessment in schools and colleges not just guides in further developing schooling assessment, advancing constant improvement of schooling objectives, and advancing logical education direction, however it likewise helps with the advancement of wise and normalized schooling the board in universities and colleges.

Coming up next are the primary developments points of this paper:

  • This review presents a clever versatile proposed hybrid classifier is formed by combining the Bi-LSTM and CNN models that can enough assess the language teaching quality and teaching impact of high school to appropriately assess the teaching quality and teaching impact.

  • As an original proposed hybrid classifier is formed by combining the Bi-LSTM and CNN models is proposed in this review. The model contains a movable learning rate and energy term to work on the hybrid classifier of artificial intelligence technique for the assembly speed and advance the network geography to guarantee the model’s stability.

Literature review

Related works

Student interest was favourably correlated with teacher competence (pedagogical topic understanding, self-efficacy, and instructional excitement), while student accomplishment was strongly correlated with self-efficacy. These associations were mediated by three aspects of teaching quality (cognitive activation, supportive atmosphere, and classroom management), which speak to the actual teacher-student interactions in the classroom [23]. To enhance instruction and all aspects of school administration, notably in institutions located in underprivileged communities. In addition, the significance of include practitioners’ viewpoints in such approaches is discussed, as are issues with instrument validity in various educational contexts and the negative effects of underperformance on the school climate [24]. We concentrate on faculty members’ views towards this change and investigate their correlations with underlying motives, burnout/engagement, and student learning in order to understand interindividual differences in online teaching and learning within this unexpected condition.

The goals of the faculty’s learning method were positively correlated with their perception of the move to online teaching as a rewarding challenge and as being beneficial to their own competence growth. However, perceptions of this transition as threatening were accompanied by performance (appearance) avoidance and work avoidance goals, which in turn were positively correlated with burnout levels and negatively correlated with student assessments of the quality of the teaching [25]. For this study, the School Teacher Effectiveness Questionnaire (STEQ), created and validated was used to assess teacher effectiveness. High reliability was found for the STEQ. These students’ English and maths test scores were also gathered from their individual schools. The association between student achievement and teacher effectiveness was calculated using Pearson correlation. The study discovered a modestly significant positive connection between student achievement and teacher effectiveness. Effective communication was found to have the second-highest correlation with student achievement in English and mathematics after the learning environment [26].

The education system that guarantees the contemporary quality of generations’ education is modelled in the context of students achieving new educational results with the aid of ICTs, management activities of teachers to create conditions for the implementation of requirements for the development of fundamental educational programmes of general education in the conditions of operating of the school information and educational environment [27]. Documentation and questionnaires were employed to collect the data, and descriptive statistical analysis methods as well as multiple regression infrastructural statistical analysis methods with SPSS were used for the study. The findings demonstrated that, when teacher labour productivity was used as a moderating variable, educational operational costs had no impact on the production of education in high schools in the Pati district or the quality of that output. One variable, teacher work productivity, cannot be employed as a moderating factor or as an independent factor that affects educational output variables [28].

The applicability of teaching machine learning, as well as the pedagogical methods, tactics, and sustainability in relation to actual implementation in educational settings. The findings point to the necessity for pre-service teachers to receive training on how to use current machine learning introduction tools. To include machine learning into regular education, teachers should also be involved in the co-designing of materials that take into account contextual elements and, most importantly, the curriculum. Engaging educators in the development process would contextualise machine learning, resulting in social improvements with a genuine impact [29]. To evaluate the effectiveness of the e-learning approach for teaching high school-aged people mathematics in comparison to the conventional expository method. The research that was conducted was quantitative, descriptive, and correlative. With a control group and an experimental group, the research design is quasi-experimental. The outcomes demonstrate a beneficial impact of e-learning on motivation, autonomy, involvement, mathematical ideas, outcomes, and grades. Given a comparison between the e-learning approach and the expository approach, it can be said that adult students studying mathematics at the high school level benefit from using the e-learning approach [30].

The effects of COVID-19 on education and the health of educators, parents, and students by outlining the difficulties associated with distance learning online. 13 parents and 11 teachers from pre-university public institutions in 14 municipalities throughout seven regions of Kosovo participated in semi-structured interviews. According to the study’s findings, COVID-19’s effects on evaluation, distant learning inadequacies, and student overload have raised a lot of concerns among students, parents, and teachers. Opportunities to improve education quality, teacher, parent, and family assistance, along with helpful advice for those working in the field of education are also included [31]. The challenges of running middle and high school online physical education classes for the first time included: (1) the monotony of the classes within their constrained environmental conditions and constrained educational content that did not adequately convey the value of physical education; (2) trial-and-error methods applied nationwide due to a lack of expertise in running online physical education classes; and (3) very constrained evaluation guidelines proposed. Changes in strategic learning approaches are required to comprehend online physical education features and, as a result, effectively communicate the benefits of physical education. This will help address the challenges that have been identified and promote the efficient operation of online physical education classes [32]. Convolutional Neural Network with Bidirectional LSTM, Initial Input, Output (Data). In order to extract features, fewer resources are needed to represent a large collection of data. It is essentially used in this way to increase system execution. The current edge differentiate technique will feature a spot that recognises data characteristics in addition to pooling activity used to reduce over-burden or CNN results.

Methods and methodology

As per the customized learning resources, a portrayal of the problem is suggested for the exploration analysis. This paper develops the proposed model which is hybrid classifier that is figured out by combining the Bi-LSTM and CNN models. The proposed method of Deep Learning approach is depicted in Fig. 1.

Fig. 1
figure 1

Proposed model

Making or selecting a data set for the information revelation’s use is included in the data selection process. This procedure has a database as its source, and the output is the target data. Data cleaning and commotion eradication pre-processing tasks are carried out. Additionally, necessary information is acquired to display or record for commotion, and appropriate procedures are established for managing missing information and expressing repetitious information. Based on the data mining task, the pre-processed data should be transformed into a preset design. To address the data in this sequence, the right kind of highlights must be chosen. At this point, aspect reduction can also be accomplished by using highlight selection. The last step is a collection of highlights that are seen as a data set. A hybrid classifier will be created in the characterisation stage and trained using the selected highlights (ILDA). The data evaluates the link between a single component and the target class and shows the strength of the association between two irregular attributes or elements, which reduces the overt repetition of element features. In order to assist sentence-supporting LSTM from now on and in response to demand, a specific ultimate purpose to kind use of mutually the previous and forthcoming set data of a decree in predicting word.

Data selection

Data selection step includes making or choosing a data set to be performed by the information revelation. The source of this process is a database and result is a target data.

Pre-processing

The Pre-processing activities for data cleaning and commotion eliminating are performed. Required data is likewise gathered to display or record for commotion and proper methodologies are determined for managing missing data and representing repetitive data.

Data transformation

The pre-processed data should be changed into a predefined design, contingent upon the data mining task. This progression requires the selection of appropriate kinds of highlights to address the data. Highlight selection can likewise be utilized at this stage for aspect decrease. The last process is a bunch of highlights perceived as a data set.

Feature selection mode

The properties of differentiated attributes is learning resources. There are many variables that influence learners’ decision of learning resources, which might incorporate qualities like orientation, generals, erudition objectives, content inclinations, erudition styles, mental levels, and learning inspirations; then again, learning resources might have inborn traits, for example, resource styles and cooperation strategies. In this manner, it is important to track down the connection among students and possessions among many highlights and lay out a component selection model to finish the info development of the proposal technique. The sifting highlight selection strategy can generally straightforwardly utilize the exhibition of the training data to assess the elements, and it doesn’t have anything to do with the ensuing procedure; the speed is quicker, and the connection amidst the real data is chosen, so the technique in view of Improved Linear Discriminant Analysis (ILDA) is chosen. Then, another hybrid classifier will be built in the characterization stage, which will be trained with the chose highlights (ILDA). The data portrays the strength of the relationship amidst two irregular ascribes or elements and judges the connection amidst’s a solitary component and the target class, consequently diminishing the overt repetitiveness of element aspects.

Optimization model

Optimization strategies are utilized to distinguish examples to anticipate the following occasion given the accessible info. In the Bi-LSTM and (ConvNet/CNN) models, the forecast alludes to prompting models that can foresee whether an understudy successful learning (Hubscher et al. 2007). Most forecast strategies depend on straightforward measurable model like relapse, nonlinear insights and neural networks. The expectation is regulated learning task where the data is utilized straightforwardly to foresee the class worth of another example. This strategy is utilized to track down obscure or missing qualities. Expectation includes two fundamental purposes inside CNN. Expectation strategy is utilized to concentrate on the network of models which is huge for forecast giving data about the fundamental build. This is a typical methodology in projects of exploration that endeavor to foresee understudy educational results without anticipating in the middle or refereeing factors first. To observe the result esteem in outline works in expectation, the subsequent strategy is utilized. For instance, in recently gathered storehouse data, where wanted marked data may not be accessible, or in settings where getting names could change the way of behaving being named.

LSTM-long short-term memory

LSTM regards are not adjusted as erudition proceeds. Neural Networks grant headlong and in switch relationship amidst neurons. To arrange the method and expect while plan is accomplish by LSTM. Comparative mercilessness to opening provides great situation to LSTM over choice RNNs, covered Markov representations and other game plan learning systems in different claims. Intermittent optimizations are capable to getting long-eliminate environments, but they bomb as a result of the point evaporating/exploding matters. LSTMs is familiarize for with deal with RNNs tendency disappearing problems.

Bi-directional LSTM

Particular ultimate objective to kind use of mutually the previous and forthcoming set data of a decree in anticipating word, along these lines, bidirectional prototypical by supporting sentence to LSTM from onward and in turn around demand. The Structural Method of LSTM implementation is shown in Fig. 2.

Fig. 2
figure 2

Bi-Directional LSTM Architecture

CNN

Convolutional neural network is a domain of deep, advanced AINN that used for convolutional and pooling reason. To decrease resources for data recovery, progression a Convolutional neural framework was available. There are four layer of CNN:

The Fig. 3 explains the four important layers of CNN Architecture. At first, CNN attractive system made data as an info. Then, at that point, its show out a deuce activity. Termed by Convolution and Pooling. The board of these activities is to acquire unalike spatial data of values. In any case the convolution activity completes 2D involvedness. Be that as it may, the basic activity of CNN is edge acknowledgment. By then the pooling activity fills two crucial requirements. It reduces the degree of yield by decline resources from open data. Its, also perform understanding and turn of framework.

Fig. 3
figure 3

CNN Architecture

Proposed model

  • Initially, Input.

  • Convolutional Neural Network with Bidirectional LSTM.

  • Output (Data).

Feature extraction includes lessening the quantity of resources expected to depict an enormous arrangement of data. In this way, essentially its utilization for increment execution of System. Presently, edge distinguish process will have a spot that recognize traits of data lastly, pooling activity utilized for lessen over-burden or result from CNN.

The results of BI model

An application for teaching help based on BP neural network approach will be discussed briefly in this part. The MATLAB programming is acknowledged for the replication inquiry of education quality evaluation, and BiLSTM with CNN improved the mean-square-mistake variety chart of CNN. Due to this, each component of the suggested approach’s teaching support experiment must be tested separately, beginning with a single upgraded module. When it comes to True positive rate, True negative rate, false positive rate, and false negative rate, BiLSTM, CNN, and SVM are faster than other comparable algorithms. Quality assurance models for Open Education Resources. Additionally, we have found an appropriate OER dataset (the VLN dataset) that will allow us to extract the aforementioned features and create high-quality models. With regard to the prospective quality models that could be created utilising the VLN dataset. There are some limitations to the VLN dataset that we have discovered. It is a selection of lectures on video. Most lectures are given in English.

Experimental setup

The using Confusion matrix of True positive rate, True negative rate, false positive rate and false negative rate are visualized through following figures for training data of 90%, 80%, 70% and 60% respectively. The average assessment accuracy of the confusion matrix of the classification model of training data proposed model comparing with BiLSTM, CNN, SVM for 90% training data are given in Table 1. From 90% of training data the proposed model has 0.057, 0.005, 0.995 and 0.943 of FPR, FNR, TPR and TNR respectively.

Table 1 90% of training data of confusion matrix

The BiLSTM model is 0.065, and 0.853 of false positive rate and true positive rate, and the average assessment accuracy of the CNN algorithm has 0.077, and 0.842 of false positive rate and true positive rate. It can be seen that the proposed model algorithm has better assessment results. Confusion matrix of True positive rate, True negative rate, false positive rate and false negative rate are visualized through following Fig. 4(a), (b), (c), (d), (e) and (f) for training data of 90%.

Fig. 4
figure 4

Comparison of Different Optimization Techniques for 90% of Training Data

80% Of training data

The average assessment accuracy of the confusion matrix of the classification model of training data proposed model comparing with BiLSTM, CNN, SVM for 80% training data are given in Table 2. From 80% of training data the proposed model has 0.052, 0.005, 0.995, and 0.948 of FPR, FNR, TPR and TNR respectively. The BiLSTM model is 0.057, and 0.921 of false positive rate and True positive rate, and the average assessment accuracy of the CNN algorithm has 0.060, and 0.912 of false positive rate and True positive rate. Confusion matrix of True positive rate, true negative rate, false positive rate and false negative rate are visualized through following Fig. 5(a), (b), (c), (d), (e) and (f) for training data of 80%.

Table 2 80% of training data of confusion matrix
Fig. 5
figure 5

Comparison of Different Optimization Techniques for 80% of Training Data

Using 70% of the training data

The average assessment accuracy of the confusion matrix of the classification model of training data proposed model comparing with BiLSTM, CNN, SVM for 70% training data are given in Table 3. From 70% of training data the proposed model has 0.055, 0.002, 0.998 and 0.958 of FPR, FNR, TPR and TNR respectively.

Table 3 70% of training data of confusion matrix

The BiLSTM model is 0.059, and 0.941 of false positive rate and True positive rate, and the average assessment accuracy of the CNN algorithm has 0.061, and 0.922 of false positive rate and True positive rate. Confusion matrix of true positive rate, true negative rate, false positive rate and false negative rate are visualized through following Fig. 6(a), (b), (c), (d), (e) and (f) for training data of 70%.

Fig. 6
figure 6

Comparison of Different Optimization Techniques for 70% of Training Data

Using 60% of training data

A number of trials were carried out to confirm that the teaching support resources suggested by the method in this study suit the needs of learners. Along with information on learning resources, experimental data also includes learning history. It offers dozens of attributes in the public data sets that are already available, such as EDX, World UC, and other data sets, including course data, teaching quality support data, and learner behaviour data. In this experiment, the learning rate is initially set at 0.001 and decreases to one-tenth of the original 90%, 80%, 70%, and 60% training data sets after 15,000 and 20,000 repetitions. The average assessment accuracy of the confusion matrix of the classification model of training data proposed model comparing with BiLSTM, CNN, SVM for 60% training data are given in Table 4.

Table 4 60% of training data of confusion matrix

From 60% of training data the proposed model has 0.042, 0.002, 0.999 and 0.959 of FPR, FNR, TPR and TNR respectively. The BiLSTM model is 0.060, and 0.941 of false positive rate and True positive rate, and the average assessment accuracy of the CNN algorithm has 0.061, and 0.922 of false positive rate and True positive rate. Confusion matrix of True positive rate, True negative rate, false positive rate and false negative rate are visualized through following Fig. 7(a), (b), (c), (d), (e) and (f) for training data of 60%.

Fig. 7
figure 7

Comparison of Different Optimization Techniques for 60% of Training Data

Feature selection with optimization techniques

The distinction results of typical valuation precision using Feature selection optimization techniques are shown in Table 5. The typical quality accuracy of the proposed model for 100% of groups of data using the proposed model accuracy is 97%, the typical quality accuracy of the optimized Seagull optimization algorithm model is 95%, and the typical quality accuracy of the Red Colobuses Monkey algorithm is 93 respectively. It is shown that the proposed model algorithm has better quality results. The outcome portraits that the proposed hybrid classifier is formulated by combining the Bi-LSTM (Bi-directional long short-term memory) and Convolutional Neural Network (ConvNet/CNN) models show better language educational assessment estimation prototypical optimized by a deep learning algorithm based on mobilize artificial intelligence can make modernized appraisal results on educational quality. The Comparison curve of feature selection of the optimization is shown in Fig. 8.

Table 5 Comparison of feature selection with optimization techniques
Fig. 8
figure 8

Comparison curve of feature selection with optimization techniques

The proposed algorithm’s accuracy is compared to that of the recommended, SOA, RCM, PSO, and other algorithms in Table 6 using the same data set. The accuracy index of the proposed method is significantly greater than that of the other three algorithms, with the exception of all course categories. The suggested algorithm is better suited for the recommendation of courses with substantial temporal dependence than the other three techniques.

Table 6 Accuracy comparison of four algorithms

Experimental results

Several experiments were done to see if the teaching assistance resources suggested by the strategy in this study actually addressed the demands of the students. Both previous learning data and data from learning resources are included in experimental data. It offers dozens of attributes in the public data sets that are already available, including information on courses, teaching quality support, and learner behaviour. In this experiment, the learning rate starts off at 0.001 and drops to 0.010 after 15,000 and 20,000 iterations, when 90%, 80%, 70%, and 60% of the original training data sets are fed. In this trial, the normal worth acquired by running a few times is the eventual outcome. The learning pace of the model adjusts adaptively and powerfully because of changes in mistakes, guaranteeing that the model’s union speed is improved during the training process, and the blunder in the midst of the model and the normal worth of the target can be incredibly diminished. It is additionally more compelling than customary strategies in foreseeing accuracy and cycle times.

Conclusion

The estimation of language educational assessment is an extremely perplexing process, which includes many elements and factors, so the foundation of the numerical model is muddled, and the conventional assessment technique for language teaching quality is as of now not completely able. Utilizing modernized AI innovation to advance education and teaching change to meet the improvement has turned into an expansive agreement in the education local area. The proposed hybrid classifier is formed by combining the Bi-LSTM and Convolutional Neural Network (ConvNet/CNN) models algorithm built in this paper is clearly better than the expectation aftereffects of an unadulterated CNN in training rate and fitting outcomes. The explanation is that the impact and execution of the customary deep learning algorithms are enormously restricted by the irregular instatement of loads and edges, and the combination impact of the network isn’t great. By improving the neural network’s forecast accuracy and blending speed, the findings of the assessment of teaching quality were improved, resulting in a more realistic framework for evaluating the quality of language instruction in high schools. Utilising the Bi-directional Long Short Term Memory (Bi-LSTM) and Convolutional Neural Network (ConvNet/CNN) models, we have directed simulated experiments and comparison analyses. Then, using hybrid optimising model parameters created by the Seagull Optimisation Algorithm (SOA) and Red Colobuses Monkey (RCM), a framework for evaluating the quality of language teaching is built out.

Data Availability

The authors do not provide supplementary data and material.

References

  1. Guo J. Empirical Analysis for English Teaching Integration and Optimization Based on Big Data Mining Technology. In Proceedings of the 2020 International Conference on Computers. Information Processing and Advanced Education, 2020, pp. 504–508. https://doi.org/10.1145/3419635.3419736

  2. Qiu C. Empirical Study of Big Data Mining Technology in English Teaching Integration and Optimization Analysis. In Proceedings of the 2020 International Conference on Computers, Information Processing and Advanced Education. 2020, pp. 495–499. https://doi.org/10.1145/3419635.3419734

  3. Zhang H, Tsai SB. An empirical study on Big Data Model and visualization of internet + teaching. Math Probl Eng. 2021. https://doi.org/10.1155/2021/9974891

    Article  Google Scholar 

  4. Hou W. Analysis of key indicators in English teaching evaluation based on Big Data Model. Sci Program. 2022. https://doi.org/10.1155/2022/1231700

    Article  Google Scholar 

  5. Duan J, Gao R. Research on college English teaching based on data mining technology. Eurasip J Wirel Commun Netw. 2021;1:1–12. https://doi.org/10.1186/s13638-021-02071-6

    Article  Google Scholar 

  6. XU J. Research on Multidimensional Teaching Mode of College English based on Data Mining. 2020 Int Conf Big Data Social Sci (ICBDSS) Xi’an China. 2020;1–4. https://doi.org/10.1109/ICBDSS51270.2020.00008

  7. Yang L, Data Mining Based College English Teaching Assistant Expert System. 2021 6th International Conference on Smart Grid and Electrical Automation (ICSGEA), Kunming, China, 2021, pp. 280–283. https://doi.org/10.1109/ICSGEA53208.2021.00068

  8. Yu L, College English Teaching. Application Research of SVM-based Mining Algorithm in Evaluation of. 2016 International Conference on Intelligent Transportation, Big Data & Smart City (ICITBS), Changsha, China, 2016, pp. 73–76. https://doi.org/10.1109/ICITBS.2016.124

  9. Wu Y, Huang H. Data Mining in Teaching Quality Analysis: a Case Study in College English teaching. 2009 Int Conf Comput Intell Softw Eng Wuhan China. 2009;1–4. https://doi.org/10.1109/CISE.2009.5367040

  10. Wang J, Research on College English Teaching Strategies and Applications Based on Big Data. 2019 International Conference on Machine Learning, Big Data and Business Intelligence (MLBDBI), Taiyuan, China, 2019, pp. 283–287. https://doi.org/10.1109/MLBDBI48998.2019.00063

  11. Wan L. Extraction Algorithm of English Text Summarization for English Teaching, 2018 International Conference on Intelligent Transportation. Big Data & Smart City (ICITBS), Xiamen, China, 2018, pp. 307–310. https://doi.org/10.1109/ICITBS.2018.00085

  12. Kong H, Shu Y, Shi P, Application Research of Personalized Recommendation Technology in College English Teaching Reform under The Background of Big Data. 2021 6th International Conference on Smart Grid and Electrical Automation (ICSGEA), Kunming, China, 2021, pp. 468–472. https://doi.org/10.1109/ICSGEA53208.2021.00112

  13. Wu HA, Method for Classification of English Reading Materials Based on Web Information Mining. 2009 First International Workshop on Education Technology and Computer Science, Wuhan, China, 2009, pp. 465–468. https://doi.org/10.1109/ETCS.2009.112

  14. Sun M, Li Y. Eco-Environment Construction of English Teaching Using Artificial Intelligence Under Big Data Environment, in IEEE Access, 8, 2020, pp. 193955–193965. https://doi.org/10.1109/ACCESS.2020.3033068

  15. Liang J, Design of Evaluation Model of English Teaching Achievement in Microcourse. 2018 10th International Conference on Measuring Technology and Mechatronics Automation (ICMTMA), Changsha, China, 2018, pp. 497–500. https://doi.org/10.1109/ICMTMA.2018.00126

  16. Xue L, Aidong G. The application of data mining technology in the college English network self-learning monitoring system. 2013 3rd International Conference on Consumer Electronics Communications and Networks Xianning China. 2013;666–8. https://doi.org/10.1109/CECNet.2013.6703418

  17. YANG Y, WANG C, Analysis of Business English Information Mining Method Based on Task Cooperative Learning Model. 2019 International Conference on Robots & Intelligent System (ICRIS), Haikou, China, 2019, pp. 344–348. https://doi.org/10.1109/ICRIS.2019.00093

  18. Xiaoya G, Kan L, Ping L. Visual analysis of college students’ scores in English test. 2009 4th International Conference on Computer Science & Education Nanning. 2009;1816–9. https://doi.org/10.1109/ICCSE.2009.5228253

  19. Wu J. Empirical Analysis of Evaluation of English Teachers’ Educational Ability under MOOC Environment. 2018 International Conference on Intelligent Transportation, Big Data & Smart City (ICITBS), Xiamen, China, 2018, pp. 303–306. https://doi.org/10.1109/ICITBS.2018.00084

  20. Chen D, Aghdam AR, Kamalpour M, Sim ATH. The impact of College English Test (CET) on graduates’ salaries using data mining techniques. 2013 International Conference on Research and Innovation in Information Systems (ICRIIS), Kuala Lumpur, Malaysia, 2013, pp. 559–563. https://doi.org/10.1109/ICRIIS.2013.6716770

  21. Yan Q, Li D, English Teaching Analysis Algorithm Design Based on Bidirectional Association Rules. 2021 IEEE Asia-Pacific Conference on Image Processing, Electronics and Computers (IPEC), Dalian, China, 2021, pp. 1117–1121. https://doi.org/10.1109/IPEC51340.2021.9421129

  22. Zhang J, Development of Smart English Guiding based on Internet Data Analysis under the Background of Internet. 2021 Second International Conference on Electronics and Sustainable Communication Systems (ICESC), Coimbatore, India, 2021, pp. 738–741. https://doi.org/10.1109/ICESC51422.2021.9532620

  23. Fauth B, Decristan J, Decker A-T, Büttner G, Hardy I, Klieme E, Mareike Kunter. The effects of teacher competence on student outcomes in elementary science education: the mediating role of teaching quality. Teach Teacher Educ. 2019;86:102882.

    Article  Google Scholar 

  24. Skedsmo G. Top-down and bottom-up approaches to improve educational quality: their intended and unintended consequences. Educational Assess Evaluation Account. 2019;31:1–4.

    Article  Google Scholar 

  25. Daumiller M, Rinas R, Hein J, Janke S, Dickhäuser O, Dresel M. Shifting from face-to-face to online teaching during COVID-19: the role of university faculty achievement goals for attitudes towards this sudden change, and their relevance for burnout/engagement and student evaluations of teaching quality. Comput Hum Behav. 2021;118:106677.

    Article  Google Scholar 

  26. Akram M. Relationship between students’ perceptions of teacher effectiveness and student achievement at secondary School Level. Bull Educ Res. 2019;41(2):93–108.

    Google Scholar 

  27. Musokhonovna KL. ICT-As a means of achieving new educational results in teaching natural disciplines in secondary schools. ACADEMICIA: An International Multidisciplinary Research Journal. 2021;11(10):315–21.

    Google Scholar 

  28. Surur M, Wibawa RP, Jaya F, Suparto AA, Harefa D, Faidi A, Triwahyuni E, Kadek Suartama I. Abdul Mufid, and Agus Purwanto. “Effect of education operational cost on the Education Quality with the School Productivity as moderating variable. Psychol Educ. 2020;57(9):1196–205.

    Google Scholar 

  29. Sanusi I, Temitayo SS, Oyelere, Joseph Olamide Omidiora. Exploring teachers’ preconceptions of teaching machine learning in high school: a preliminary insight from Africa. Computers and Education Open. 2022;3:100072.

    Article  Google Scholar 

  30. Moreno-Guerrero. Antonio-José, Inmaculada Aznar-Díaz, Pilar Cáceres-Reche, and Santiago Alonso-García. “E-learning in the teaching of mathematics: An educational experience in adult high school.“ Mathematics 8, no. 5 (2020): 840.

  31. Duraku ZH, Hoxha L. “The impact of COVID-19 on education and on the well-being of teachers, parents, and students: Challenges related to remote (online) learning and opportunities for advancing the quality of education.“ Manuscript submitted for publication]. Faculty of Philosophy, University of Prishtina (2020).

  32. Jeong H-C, Wi-Young S. Difficulties of online physical education classes in middle and high school and an efficient operation plan to address them. Int J Environ Res Public Health. 2020;17(19):7279.

    Article  Google Scholar 

Download references

Acknowledgements

None.

Funding

Not applicable.

Author information

Authors and Affiliations

Authors

Contributions

L.C. is the first author role, he performed the literature review, implemented the proposed model, carried out the experiments and wrote the manuscript. M.M. has a supervisory role, she oversaw the completion of the work. Both authors read and approved the final manuscript.

Corresponding author

Correspondence to Mazlin Mohamed Mokhtar.

Ethics declarations

Consent for publication

Not applicable.

Disclosure of potential conflict of interest

The authors declare that they have no potential conflict of interest.

Ethical approval

All applicable institutional and/or national guidelines for the care and use of animals were followed.

Informed consent

For this type of analysis formal consent is not needed.

Additional information

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and Permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Chen, L., Mohamed Mokhtar, M. Education on quality assurance and assessment in teaching quality of high school instructors. J Big Data 10, 142 (2023). https://doi.org/10.1186/s40537-023-00811-7

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1186/s40537-023-00811-7

Keywords

  • Language Teaching
  • Bi-directional long short term memory
  • Convolutional Neural Network
  • Deep Learning
  • Teaching quality optimization