1. bookAHEAD OF PRINT
Informacje o czasopiśmie
License
Format
Czasopismo
eISSN
2444-8656
Pierwsze wydanie
01 Jan 2016
Częstotliwość wydawania
2 razy w roku
Języki
Angielski
Otwarty dostęp

Mathematical Statistics Technology in the Educational Grading System of Preschool Students

Data publikacji: 15 Jul 2022
Tom & Zeszyt: AHEAD OF PRINT
Zakres stron: -
Otrzymano: 09 Feb 2022
Przyjęty: 09 Apr 2022
Informacje o czasopiśmie
License
Format
Czasopismo
eISSN
2444-8656
Pierwsze wydanie
01 Jan 2016
Częstotliwość wydawania
2 razy w roku
Języki
Angielski
Introduction

In October 2012, the Ministry of Education promulgated the “Guidelines for the Study and Development of Children from 3 to 6 Years Old” and proposed specific indicators for promoting young children's physical and mental health. The healthy growth of children's bodies and minds is inseparable from cultivating children's emotional ability. Emphasizes the importance of children's emotional abilities for their growth and success. Studies have pointed out that only 20% of life achievements are attributed to IQ, and the other 80% are affected by other factors. People with higher emotional abilities are dominant at all stages of life and have more chances of success [1]. However, today's education system mostly emphasizes the development of children's intelligence but generally ignores children's emotional abilities. Emotional ability can be improved by training. Early childhood is a critical period for the development of emotional, cognitive ability. The traditional way of teaching emotions is positive and effective in cultivating children's emotion recognition ability. However, with the continuous advancement of educational technology, we need to apply new technologies to the training and training of children's emotion recognition ability to reduce costs further and enhance interest. This article uses Android mobile devices as teaching methods to design scene animation and voice systems. The system sets up interesting storylines to realize children's immersive situational teaching method as the main body. We tested and analyzed the effectiveness of this system.

The design and application of the teaching system of children's emotion cognition based on APP Inventor
System function design

The system needs to have the following main functions: First, it can answer questions about children's emotions [2]. The second has the function of displaying the answer results. Third, it has a fun game function of level-breaking level. The fourth question is the plot of the story. The functional structure of the children's emotional cognition teaching system based on APP Inventor is shown in Figure 1.

Figure 1

System functional structure

Emotion question and answer

In this module, the questions are described and conveyed to the children in text, voice, and scene pictures. Set 3 options for each question in this module of options. Each option sets emotional expression pictures and attaches text descriptions.

Answering results

This module consists of two sub-modules, correct and error. In this module, correct and wrong judgments are made on the answer of the respondent. This design can promote children's thinking and encourage them to recognize emotions to achieve the purpose of emotion teaching.

Level of breakthrough

This module is the center of the system. It consists of three sub-modules: simple, common, and difficult. This module runs through the entire infant emotional test system. We divide the system into 3 sub-levels. Each level is composed of 3 questions, and the questions in the 3 levels range from simple to difficult. After the answerer completes a level question, it will automatically jump to the reward interface [3]. At the same time, the user can choose to continue to the next level or not to continue to the next level and stop the game.

The realization of the teaching system of children's emotional cognition

We use APP Inventor as the basic development platform to implement children's emotion cognition teaching system from the component and logic layers.

Component layer design

The component layer is the interface of the children's emotional cognition teaching system. It is the operation interface and facade of the children's emotional cognition teaching system based on APP Inventor. The design of the component layer determines the external operation mode of the system [4]. The component layer is divided into visual and non-visual components. A visual component is a component type that the human eye can directly observe on the screen. In contrast, a non-visual component is a component that is not directly observed by the human eye. The content includes components such as audio and video players.

Logic layer design

The components and logic layers can be switched as needed in the children's emotional cognition teaching system. The “code blocks” encapsulated in the logic layer are stored in “drawers” designed in different building layers. We select module splicing from the drawer and assemble the program. The emotional question-and-answer session is the core part of the entire system. The logic design of the emotional question-and-answer session is divided into three parts, namely, the setting of the question cycle, the setting of the right or wrong judgment of the options, and the setting of animation. The question loop setting can make three different question types with the same difficulty appear on one answer screen, and its global variables and definition process are initialized. Judging whether the option is right or wrong is mainly the setting of the options button and the call to the non-visual component [5]. The animation set is mainly realized through the visual component web viewer and the sound button. In the whole process of the children's emotional cognition teaching system, computational thinking is used to achieve, and the system design is continuously improved through the project team. We decompose system functions into different modules. And we use a modular approach to achieve the different functional system implementation processes of the complete system (as shown in Figure 2).

Figure 2

System implementation process

Based on multi-scale and improved children's emotional mechanism model
Model establishment

The coding and decoding model of infant emotion mechanism mainly includes three modules. The encoder is responsible for processing the input sequence x. We denote it as hE=[h1E,h2E,,hSE]RS×N {h^E} = \left[ {h_1^E,\,h_2^E,\, \cdots ,\,h_S^E} \right] \in {R^{S \times N}} and provide it to the decoder. The infant emotion module is a structure extended from the classic model [6]. We are used to helping the decoder find more relevant information from the hidden layer. At time t, the infant emotional module processes the contextual information ct through equation (1). ct=s=1Sat[s]×hsEat[s]=Align(hsE,htD)=exp(Score(hsE,htD))s=1Sexp(Score(hsE,htD)) \matrix{ {{c_t} = \sum\limits_{s = 1}^S {{a_t}\left[ s \right] \times h_s^E} } \hfill \cr {{a_t}\left[ s \right] = \,Align\left( {h_s^E,\,h_t^D} \right) = {{\exp \left( {Score\left( {h_s^E,\,h_t^D} \right)} \right)} \over {\sum\limits_{s = 1}^S {\exp \,\left( {Score\left( {h_s^E,\,h_t^D} \right)} \right)} }}} \hfill \cr }

Align is the alignment function in the children's emotional mechanism. Score represents the mapping and (RM × RN) → R, M represents the number of hidden units of the encoder. N represents the number of hidden units of the decoder. The decoder module finally generates the target sequence by formula (2) according to the output of the previous layer and the context information ct at time t. logP(y|x;θ)=t=1TlogP(yt|htD,ct;θ) \log \,P\left( {y\left| {x;\theta } \right.} \right) = \sum\limits_{t = 1}^T {\log \,P\left( {{y_t}\left| {h_t^D,\,{c_t};\theta } \right.} \right)}

In the formula: htD h_t^D represents the last decoding layer, which contains all the summary information input by the y < t layer; θ is the parameter of the model.

Emotional mechanism of young children

The scoring equation of the infant's emotional mechanism generates a score to calculate the correlation between the source and target sides. The emotional mechanism of young children can be modeled as dot product, bilinear product, and multi-layer perceptron (MLP). The scoring equation is as follows: Score(hsE,htD)=m=1MhsE[m]×htD[m] Score\left( {h_s^E,\,h_t^D} \right) = \sum\limits_{m = 1}^M {h_s^E\left[ m \right] \times h_t^D\left[ m \right]} htD h_t^D and hsERM h_s^E \in {R^M} . There are no trainable parameters in equation (3). The scoring equation of the bilinear product model is as follows: Score(hsE,htD)=hsEWhhtD Score\left( {h_s^E,\,h_t^D} \right) = h_s^E\,Whh_t^D hsERM h_s^E \in {R^M} , htDRN h_t^D \in {R^N} , WRM×N is a model parameter that can be trained. The scoring equation of MLP mode is as follows: Score(hsE,htD)=W3tanh(W1hsE+W2htD) Score\left( {h_s^E,\,h_t^D} \right) = {W_3}\,\tanh \left( {{W_1}h_s^E + {W_2}h_t^D} \right) hsERM h_s^E \in {R^M} , htDRN h_t^D \in {R^N} , W1RP×N, W2RP×N and W3RP are the model parameters that can be trained. The result of the above scoring function score is a non-standardized scalar value [7].

Sequence-to-sequence automatic speech recognition

We apply the sequence-to-sequence model to automatic speech recognition (ASR). Suppose x is a sequence of a series of audio features, xRS×F, where F is the number of features. S is the total frame length of the sound. The output y is a speech transcription sequence, a phoneme sequence, or a character record sequence. The sequence-to-sequence ASR principle based on the emotional mechanism of young children is shown in Figure 3.

Figure 3

The principle of sequence-to-sequence ASR based on the emotional mechanism of young children

Sequence-to-sequence speech synthesis

We apply the sequence-to-sequence model to speech synthesis (TTS). The model can be optimized in the training phase by minimizing the loss function shown in equation (6). LossTTS(x,x^,b,b^)=s=1SxxMx^xM2+xsRx^sR2(bslog(b^s)+(1bs)log(1b^s)) \matrix{ {Los{s_{TTS}}\left( {x,\,\hat x,\,b,\,\hat b} \right) = \sum\limits_{s = 1}^S {{{\left\| {x_x^M - \hat x_x^M} \right\|}^2} + {{\left\| {x_s^R - \hat x_s^R} \right\|}^2} - } } \hfill \cr {\left( {{b_s}\,\log \left( {{{\hat b}_s}} \right) + \left( {1 - {b_s}} \right)\log \left( {1 - {{\hat b}_s}} \right)} \right)} \hfill \cr } x^sM \hat x_s^M is the predicted logarithmic mel spectrum. x^sR \hat x_s^R is the predicted log-linear spectrum parameter. b^s {\hat b_s} is the predicted final frame probability. x^sM \hat x_s^M , x^sR \hat x_s^R and bs are real data. The sequence-to-sequence TTS principle is shown in Figure 4.

Figure 4

Sequence-to-Sequence TTS principle based on young children's emotional mechanism

Effectiveness test and analysis of the teaching system of children's emotion cognition
Subject

All the children collected in the data are of the same age and in good physical condition. We used a random sampling method based on gender classification to select 12 children aged 4 to 5 as the subjects [8]. We organized the young children into the experimental group and the control group.

Experimental method
Pre-test test

The pre-test test refers to the non-teaching test. One trained kindergarten teacher was selected as the main text of the experimental group and the control group and was responsible for the test of the subjects in the group. The main tester of the experimental group opened the software and listened to stories and pictures from the children. The control group showed children pictures and read stories. The two groups did not explain and finally let the children answer emotional questions [9]. Each question is answered correctly, and 1 point is counted, and incorrect answers are not scored.

Post-test experiment

The experimental group used “mobile phones,” The control group used “paper picture books.” Regardless of whether the answer to the pre-test is correct or not, the test teacher will teach each question of the two groups of children for 3 minutes and then conduct the test. We record post-test results [10]. Each question is answered correctly, and 1 point is counted, and incorrect answers are not scored.

Data collation
Pre-test results

The pre-test score refers to the test scores of children's emotional theme comprehension ability in the experimental group using “mobile phones” and the control group using “paper picture books,” respectively. The test results of 6 children in the experimental group and the control group are shown in Table 1.

Pre-test scores of the experimental group and the control group.

Experiment Test group Control group
Male 1 3 2
Male 2 3 4
Male 3 3 3
Female 1 4 2
Female 2 2 2
Female 3 1 3
Post-test results

The post-test results refer to the test group using “mobile phones” and the control group using “paper picture books,” respectively. We first carry out 3 minutes of full auxiliary teaching for each question and then the test scores of children's emotional theme comprehension ability [11]. The test results of 6 children in the experimental group and the control group are shown in Table 2.

Post-test results of experimental group and control group.

Experiment Test group Control group
Male 1 9 8
Male 2 9 7
Male 3 7 9
Female 1 8 9
Female 2 6 7
Female 3 9 7
Overall system test

We use the LJSpeech data set in the part of the system effectiveness experiment. Among them, 94% of the data is used as training data, and 3% is used as test data [12]. We tested the speech error rate (WER) of young children generated by this method and the control method. The experimental results are shown in Table 3.

Comparison of system effectiveness wer.

Method WER/%
Basic MLP 9.44
MLP + Location Aware 7.91
NOSPLC 7.96
High GRU 6.14
Text model + multi-scale aggregation 5.02
Text model + historical context 4.38
Text model + multi-scale and context 3.96

In the system effectiveness experiment, the basic multi-layer perceptron model, multi-layer perceptron model + Location Aware, and high GKU are still used as control methods. We analyze the time cost of various methods under the same data scale. The experimental results are shown in Table 4.

System effectiveness time overhead.

Method 25% 50% 75%
Basic MLP 13.77 29.34 45.05
MLP + Locate Aware 16.92 34.15 51.25
NOSPLC 16.53 32.06 51.43
Light GRU 14.79 29.50 45.47
Text model + multi-scale aggregation 14.07 29.53 45.31
Text Model + History Previous and Next 14.34 29.59 46.02
Text model + multi-scale and upper and lower text 15.01 30.47 46.95

It can be seen that when the test data are 25%, 50%, and 75% of the total data set size, the time cost of the method in the three modes is second only to the basic MLP method. This method is better than the other two comparison methods.

Conclusion

The “mobile phone” application system of the children's emotional, cognitive teaching software based on APP Inventor can significantly improve children's ability to understand emotional topics. With the growth of the data scale, the time cost of the method in this paper grows more slowly and has better time performance.

Figure 1

System functional structure
System functional structure

Figure 2

System implementation process
System implementation process

Figure 3

The principle of sequence-to-sequence ASR based on the emotional mechanism of young children
The principle of sequence-to-sequence ASR based on the emotional mechanism of young children

Figure 4

Sequence-to-Sequence TTS principle based on young children's emotional mechanism
Sequence-to-Sequence TTS principle based on young children's emotional mechanism

System effectiveness time overhead.

Method 25% 50% 75%
Basic MLP 13.77 29.34 45.05
MLP + Locate Aware 16.92 34.15 51.25
NOSPLC 16.53 32.06 51.43
Light GRU 14.79 29.50 45.47
Text model + multi-scale aggregation 14.07 29.53 45.31
Text Model + History Previous and Next 14.34 29.59 46.02
Text model + multi-scale and upper and lower text 15.01 30.47 46.95

Post-test results of experimental group and control group.

Experiment Test group Control group
Male 1 9 8
Male 2 9 7
Male 3 7 9
Female 1 8 9
Female 2 6 7
Female 3 9 7

Pre-test scores of the experimental group and the control group.

Experiment Test group Control group
Male 1 3 2
Male 2 3 4
Male 3 3 3
Female 1 4 2
Female 2 2 2
Female 3 1 3

Comparison of system effectiveness wer.

Method WER/%
Basic MLP 9.44
MLP + Location Aware 7.91
NOSPLC 7.96
High GRU 6.14
Text model + multi-scale aggregation 5.02
Text model + historical context 4.38
Text model + multi-scale and context 3.96

Rimfeld, K., Malanchini, M., Hannigan, L. J., Dale, P. S., Allen, R., Hart, S. A., & Plomin, R. Teacher assessments during compulsory education are as reliable, stable and heritable as standardized test scores. Journal of Child Psychology and Psychiatry., 2019; 60(12): 1278–1288 RimfeldK. MalanchiniM. HanniganL. J. DaleP. S. AllenR. HartS. A. PlominR. Teacher assessments during compulsory education are as reliable, stable and heritable as standardized test scores Journal of Child Psychology and Psychiatry 2019 60 12 1278 1288 10.1111/jcpp.13070684874931079420 Search in Google Scholar

Ezer, P., Kerr, L., Fisher, C. M., Heywood, W., & Lucke, J. Australian students’ experiences of sexuality education at school. Sex Education., 2019; 19(5): 597–613 EzerP. KerrL. FisherC. M. HeywoodW. LuckeJ. Australian students’ experiences of sexuality education at school Sex Education 2019 19 5 597 613 10.1080/14681811.2019.1566896 Search in Google Scholar

Spencer, T. D., Petersen, D. B., Restrepo, M. A., Thompson, M., & Gutierrez Arvizu, M. N. The effect of Spanish and English narrative intervention on the language skills of young dual language learners. Topics in Early Childhood Special Education., 2019; 38(4): 204–219 SpencerT. D. PetersenD. B. RestrepoM. A. ThompsonM. Gutierrez ArvizuM. N. The effect of Spanish and English narrative intervention on the language skills of young dual language learners Topics in Early Childhood Special Education 2019 38 4 204 219 10.1177/0271121418779439 Search in Google Scholar

WALİD, A., SAJİDAN, S., RAMLİ, M., & KUSUMAH, R. G. T. Construction of the assessment concept to measure students' high order thinking skills. Journal for the Education of Gifted Young Scientists., 2019; 7(2): 237–251 WALİDA. SAJİDANS. RAMLİM. KUSUMAHR. G. T. Construction of the assessment concept to measure students' high order thinking skills Journal for the Education of Gifted Young Scientists 2019 7 2 237 251 10.17478/jegys.528180 Search in Google Scholar

Hu, X., Li, J. & Aram, Research on style control in planning and designing small towns. Applied Mathematics and Nonlinear Sciences., 2021; 6(1): 57–64 HuX. LiJ. Aram Research on style control in planning and designing small towns Applied Mathematics and Nonlinear Sciences 2021 6 1 57 64 10.2478/amns.2020.2.00077 Search in Google Scholar

Vanli, A., Ünal, I. & Özdemir, D. Normal complex contact metric manifolds admitting a semi symmetric metric connection. Applied Mathematics and Nonlinear Sciences., 2020; 5(2): 49–66 VanliA. ÜnalI. ÖzdemirD. Normal complex contact metric manifolds admitting a semi symmetric metric connection Applied Mathematics and Nonlinear Sciences 2020 5 2 49 66 10.2478/amns.2020.2.00013 Search in Google Scholar

York, S., Lavi, R., Dori, Y. J., & Orgill, M. Applications of systems thinking in STEM education. Journal of Chemical Education., 2019; 96(12): 2742–2751 YorkS. LaviR. DoriY. J. OrgillM. Applications of systems thinking in STEM education Journal of Chemical Education 2019 96 12 2742 2751 10.1021/acs.jchemed.9b00261 Search in Google Scholar

Suhono, S., & Sari, D. A. Developing Students’ Worksheet Based Educational Comic for Eleventh Grade of Vocational High School Agriculture. Anglophile Journal., 2020; 1(1): 29–40 SuhonoS. SariD. A. Developing Students’ Worksheet Based Educational Comic for Eleventh Grade of Vocational High School Agriculture Anglophile Journal 2020 1 1 29 40 10.51278/anglophile.v1i1.78 Search in Google Scholar

Moodie, N., Maxwell, J., & Rudolph, S. The impact of racism on the schooling experiences of Aboriginal and Torres Strait Islander students: A systematic review. The Australian Educational Researcher., 2019;46(2): 273–295 MoodieN. MaxwellJ. RudolphS. The impact of racism on the schooling experiences of Aboriginal and Torres Strait Islander students: A systematic review The Australian Educational Researcher 2019 46 2 273 295 10.1007/s13384-019-00312-8 Search in Google Scholar

Rosenbaum, J. Educational and criminal justice outcomes 12 years after school suspension. Youth & Society., 2020; 52(4): 515–547 RosenbaumJ. Educational and criminal justice outcomes 12 years after school suspension Youth & Society 2020 52 4 515 547 10.1177/0044118X17752208728884932528191 Search in Google Scholar

Havik, T., & Westergård, E. Do teachers matter? Students’ perceptions of classroom interactions and student engagement. Scandinavian Journal of Educational Research., 2020; 64(4): 488–507 HavikT. WestergårdE. Do teachers matter? Students’ perceptions of classroom interactions and student engagement Scandinavian Journal of Educational Research 2020 64 4 488 507 10.1080/00313831.2019.1577754 Search in Google Scholar

Nagovitsyn, R. S., Bartosh, D. K., Ratsimor, A. Y., & Neverova, N. V. Modernization of Regional Continuing Pedagogical Education in the «School-College-Institute». European journal of contemporary education., 2019; 8(1): 144–156 NagovitsynR. S. BartoshD. K. RatsimorA. Y. NeverovaN. V. Modernization of Regional Continuing Pedagogical Education in the «School-College-Institute» European journal of contemporary education 2019 8 1 144 156 Search in Google Scholar

Polecane artykuły z Trend MD

Zaplanuj zdalną konferencję ze Sciendo