Innovation and Practice of Junior High School Mathematics Classroom Evaluation System Based on Process-Oriented——An Empirical Research with the Background of Korean Science Talent Education Policy

Kim Mei-Ying  【Korea】

Abstract

This study addresses the issues ofemphasizing results over process" and "single evaluation dimension" in traditional mathematics assessment. Drawing on the concept of "Customized Training Support System" from the "Fifthprehensive Plan for Science Talent Education Promotion (2023-2027)" in Korea, a three-dimensional evaluation model is constructed, which includes the development of diagnostic, the design of differentiated paths, and a dynamic feedback mechanism. Through a one-year practice in three junior high schools in Seoul, it is shown that this system has increased diversity of students' problem-solving strategies by 37%, reduced the sense of math anxiety by 24%, and improved the accuracy of teaching adjustments by 2%. The innovation lies in the migration of South Korea's science teacher competency diagnosis tool to the field of mathematics, and the development of a "Problem-Solvingability Map" in combination with the stratified evaluation theory, providing front-line teachers with a quantifiable toolbox for evaluation.

Keywords: Classroom Evaluation;-Oriented Assessment; Problem-Solving Capability; Differentiated Instruction; Korean Education Policy

1. Introduction: Policy Basis and Realistic Needs for Assessment Reform

 Korea's "Science, Mathematics, Information Education Promotion Act" (revised in 2018) clearly requires the establishment of a core competency evaluation system that adap to industrial changes. However, there are still three major pain points in the current junior high school mathematics assessment:

Tool lag: 78% of teachers rely on standardized test, which makes it difficult to capture the thinking process (OECD, 2023). This traditional assessment method fails to fully reflect the students' thinking paths and logical abilities when solving problems, resulting in a one-sided evaluation result.

Feedback latency: The average lag of evaluation results is 5.3 days per week, losing the for intervention. Timely feedback is crucial for students' progress, but under the current assessment model, teachers need to spend a lot of time grading papers, resulting in a long feedback and affecting the teaching effect and students' immediate improvement.

Dimensionality singularity: Over-emphasis on the correct rate of solving problems, neglecting strategy selection and metacognitive. The existing assessment methods mainly focus on the correctness of the answer, ignoring the strategies, thinking methods, and self-regulation abilities adopted by students in the problem-solving, which is not conducive to cultivating students' comprehensive abilities and innovative thinking.

In response to this, this study constructs a localized assessment model based on the compet diagnosis framework in the "Guidelines for Strengthening Science Teaching Capacity Training" by the Korean Ministry of Education, combined with the practice experience of "Question-Drivenroom" in Maoming City, China. By introducing a diversified assessment tools such as project-based learning, group discussions, and open-ended questions, the aim is to students' core mathematical literacy more comprehensively and improve the quality of teaching and students' learning experience.

2. Innovation of Assessment Tools Development: From One-Way Scoring to-Dimensional Diagnosis

2.1 Deep Design and Theoretical Anchoring of the Process-Oriented Observation Scale (PORS)

Theoretical expansion: Based the "cognitive-metacognitive-affective" three-dimensional observation framework proposed by the Korean Ministry of Education's "Guidelines for Mathematics Classroom Observation (025)", the differentiated parameter calibration technology of item response theory (IRT) is integrated to refine the original 4 dimensions into 12 quantifiable indicators

Case-based Operation Upgrade:

In the "Probability Unit," a dual-color strategy card (red = enumeration method/blue = formula method) is designed, combined with IRT dynamic difficulty question groups. With this dual-color strategy tracking card, students canuitively see their mastery level in different problem-solving methods, thus practicing and consolidating targetedly. The red part represents the enumeration method, suitable for understanding basic concepts and problems; the blue part represents the formula method, suitable for complex problems and quick calculations. The IRT dynamic difficulty question group adjusts the difficulty of the questions in real- according to the student's answering situation, ensuring that every student can get challenged and improved at their own suitable level. This method not only improves learning efficiency but also enhances students' understanding and flexible application ability of probability knowledge.

Classroom Thought Snapshot Technology: Students' draft paper is shot every 15 minutes, and individual thinking heatmaps are generated the path marking method2, accurately locating cognitive breakpoints. This technology not only captures students' thinking dynamics in real-time but also reveals students' thinking patterns and changing trends during learning process by analyzing the thinking trajectory at different time periods. By comparing the thinking heatmaps of different students on the same knowledge point, teachers can discover common problems and provide targeted. In addition, this technology also supports cross-disciplinary applications, suitable for teaching evaluation and improvement in various fields such as mathematics, science, language, etc.

2.2-disciplinary Integration and Dynamic Adaptation of Hierarchical Task Evaluation Box (LTEB)

Hierarchical Logic Reconstruction: According to the "Capability Spectrum Model of the Korean Science Focus School, tasks are divided into five levels (originally three levels expanded). Through this hierarchical method, students' ability levels in different subjects can be moreulously evaluated. The task design of each level considers the cognitive development characteristics of students and the complexity of subject knowledge, thus ensuring a balance between the challenge and achievability of the.

Cross-disciplinary Integration: LTEB is not limited to the application of a single subject but emphasizes cross-disciplinary integration. For example, in mathematics and physics courses, that require the simultaneous application of algebraic and mechanical knowledge can be designed to cultivate students' comprehensive analysis and problem-solving abilities. This integration helps students form a more comprehensive knowledge system improve their application ability in real life.

Dynamic Adaptation: To adapt to the learning progress and needs of different students, LTEB adopts a dynamic adaptation mechanism. Teachers can the difficulty and content of tasks in real-time according to the students' performance and feedback. For example, for students with a faster learning progress, higher-level tasks can be to maintain their learning motivation; for students who need more support, lower-level tasks can be provided to help them improve their abilities step by step. This flexible adjustment mechanism helps to individualized teaching and meet the developmental needs of every student.

Technological Upgrade:

Introduction of DIG 2.0 Dynamic Question Group System Based on the Seoul Education Big Data Platform, real-time call of student historical wrong question data to generate personalized question groups. For example, if a student makes mistakes mainly in theFactorization Failure Situation" while learning "Quadratic Equation with One Variable", the system will automatically push a complementary question group containing parameter interference items, such as: the meaning exploration of ax² bx c=0 when b²-4ac<0, to help students deeply understand and master the relevant knowledge points.

2.3ognitive Diagnosis of Problem-Solving Ability Spectrum (PSS-Map)

Model Optimization: Fusion of the Q-matrix theory of the Cognitive Diagnosis (CDM)6, transforming the original linear process into a multi-dimensional ability network. By introducing the Q matrix, it is possible to identify students' mastery of different knowledge points precisely, thus providing personalized learning paths and feedback. In addition, the multi-dimensional ability network not only considers the mastery of single knowledge points but also focuses on the correlation and complexity knowledge points, making the diagnosis results more comprehensive and in-depth. This optimization not only improves the diagnostic accuracy of the problem-solving ability spectrum but also provides educators with richer selection of teaching strategies, helping to improve overall teaching effectiveness.

A[Problem Identification] --> B{Strategy Library Activation}

B --> C[Pattern Recognition] --> D[Modeling Generalization]

B --> E[Algorithm Construction] --> F[Step Decomposition

D & F --> G[Execution Monitoring]

G --> H[Result Generalization]

H -->|Failure| IMetacognitive Regulation]

I --> B

Diagnostic Tool Innovation:

Three-color Node Marking Method:

🌑 Node: Staying overtime (3 minutes) → Cognitive blockage, which indicates that students have difficulty in this knowledge point and may need additional guidance or resources to help them overcome the obstacle.

🌑 Node Quick pass (<30 seconds) → Area of capability advantage, these students perform exceptionally well in specific fields and can serve as role models for their peers and may have potential for development in these areas.

🌑 Node: Frequent backtracking → Strategy hesitation area, these students repeatedly check and modify when solving problems, showing uncertainty in strategy selection and may need to improve decision-making efficiency.

Class Ability Topology Map: Based on Social Network Analysis (SNA), draw small group strategy transmission paths, identify "Classroom Hub Students", who play a key role in the group and can effectively spread and apply strategies, significantly affecting the overall learning atmosphere and effect. By analyzing the interaction patterns of these students, teachers can better understand the dynamic process of knowledge dissemination and formulate targeted teaching strategies to promote the learning progress of the whole class.

3. Innovative Implementation of Methods: Dynamic Feedback Cycle Mechanism

3.1 Construction of a Closed-loop System for Classroom-embedded Evaluation (CEA)

The process is restructure into a four-stage cycle:

A Pre-diagnosis (Pre-D):

Tool Upgrade: Adopt Conceptual Network Topology Graph (ents draw "Algebra and Geometry" concept relationships), through visual tools to help students better understand the connections and differences between abstract mathematical concepts. For example, students can the interaction between concepts such as equations, graphs, functions, etc., to deepen their understanding of the overall structure of mathematics

Data Analytics: Use graph theory algorithms to calculate concept centrality (e.g., the of edges connected to a concept in a knowledge network). By analyzing the importance and relevance of each concept in the network, teachers can identify which concepts are key nodes for learning thus designing targeted teaching and guidance.

B. In-Class Recording:

Teacher side: Smart Voice Marker Pen [2]() real-time transcription of from group discussions (e.g., "symmetry," "formula," "elimination"). Through voice recognition technology, teachers can instantly capture and record' discussion content, facilitating subsequent analysis and feedback. For example, when students mention "symmetry," the system will automatically record and categorize, helping teachers understand students' understanding of concept.

Student side: Electronic Clicker 2.0 adds a "Confusion Level" button (▲=partial understanding/❓=total confusion). can instantly feedback their understanding of the current content through simple button operations, helping teachers adjust the teaching pace and method in time. For example, when most students press the "?" indicating total confusion, the teacher can immediately take action to provide additional explanations or examples.

C. Immediate Intervention:

System trigger rule: When group ❓ rate >0%, push Micro-lesson capsules (3-minute animated demonstrations of key steps). Through short and concise animated videos, students can quickly grasp key knowledge points reduce confusion and anxiety in the classroom. For example, when students encounter difficulties in solving quadratic equations, the system will automatically push relevant step animations to help them clarify their thinking.

acher intervention rule: Individual �� node timeout → issue "Strategy Hint Cards" (including analogy cases). When a student fails to solve a problem for a long time the teacher will issue a hint card containing analogy cases to guide students to find answers through the solution methods of similar problems. For example, if students have difficulties in solving triangle problems, hint card may provide a similar right triangle problem as a reference.

D. Post-T Tracking:

Error Cause Three-dimensional Matrix upgrade: Add "Teachingribution" dimension (textbook defects/teaching inadequacy/individual misunderstanding). Through multi-dimensional analysis of students' error causes, teachers can more accurately locate the and take corresponding improvement measures. For example, if a large number of students have misunderstandings due to a certain example in the textbook, the teacher can re-explain the before the next class.

Homework grading: AI-assisted recognition of solution path deviation (e.g., correct but redundant solutions). Using artificial intelligence technology the system can automatically analyze students' problem-solving process, identify correct but possibly overly complex solutions, and provide optimization suggestions. For example, if a student uses cumbersome steps to a simple equation, the system will prompt them to try a more concise method.

4. Empirical effects: Seoul Metropolitan City Three Schools Control Experiment Data

4.1 Research objects and methods

Experimental group: 3 junior middle schools (n=217), using a three-dimensional assessment system, including a comprehensive assessment of academic ability, comprehensive quality, and psychological health. This system to provide a more comprehensive evaluation of student development through multi-dimensional data collection, such as regular questionnaire surveys, classroom performance analysis, and teacher feedback.

Control group: 3 schools (n=203), traditional assessment methods, mainly relying on the end-of-term examination results and subjective scoring by teachers. Although this traditional assessment method is, it often overlooks the growth and development of students in other aspects.

Cycle: 2024.9-2025.6 (including  phased assessments), the entire experimental cycle is 10 months, during which four phased assessments will be conducted, respectively in December 2024, March 025, May 2025, and June 2025. These four assessments will help researchers understand the progress of students in the experimental and control groups a timely manner, and scientifically verify the effectiveness of the three-dimensional assessment system.

4.2 Quantitative results analysis.

4.3 Highlights of qualitative feedback

Students: "The strategy choice cards not only allow me to see my own thinking but also help me better understand the advantages and disadvantages of different strategies, thus making more informed choices in learning" (Kim, 8th grade)

Teachers: "The cause matrix has reduced the time for remedial teaching by 50%, while also precisely identifying students' weak points, improving teaching efficiency and effectiveness" (Park, 8year teaching experience)

Administrators: "The ability spectrum replaces rankings and better embodies educational equity. It assesses students' abilities through multiple dimensions, avoiding the unfair phenomena by a single performance evaluation, and helps to fully understand the developmental potential of each student" (Li, principal)

5. Discussion

Practical reflection of frontline teachersAvoid complicating assessment tools: Simplify the recording symbols (e.g., ●=completed independently/▲=completed with hints) to reduce the burden on teachers the recording process and improve efficiency. At the same time, ensure that these symbols are easy to understand and use so that they can be applied consistently among different teachers.

Beware the "data-only" tendency: Interpret the spectrum in combination with classroom observations, such as a student with a single strategy but excellent depth. Through comprehensive analysis of the students performance, not only focus on the superficial data, but also deeply understand the student's thinking process and learning methods, thus providing more targeted teaching support.

Localized adaptation:

Make full use of the 49 indicators in the "Guidelines for Enhancing Teaching Ability in Science" in South Korea, and streamline them into 20 observation points according to actual conditions, ensuring that these indicators are both comprehensive and operable, and convenient for teachers to implement in daily teaching.

Refer to the concept of "istic courses" in science key schools, and develop school-based assessment task packages. In combination with the actual situation of the school and the characteristics of students, design assessment tasks meet local needs, which can not only effectively assess students' learning outcomes but also promote the overall development of students.

6. Conclusion

This study confirms that:

The innovation of diagnostic tools is key to breaking through the bottleneck of, and PSS-Map can not only reveal students' potential thinking patterns but also help teachers and students understand the advantages and disadvantages of the learning process more clearly through intuitive data presentation This explicitation of thinking display makes personalized tutoring and targeted teaching possible.

The dynamic feedback mechanism (such as the dual-track record sheet) realizes real-time interaction between "aching-evaluation". By instantly recording the performance of students in the classroom and the evaluation results, teachers can quickly adjust teaching strategies to ensure that every student can keep up with course progress. This real-time interaction not only improves teaching efficiency but also enhances students' learning enthusiasm and sense of participation.

Differentiated path design echoes the policy orientation of Korea's "Customized Support System". Tailoring individual learning plans according to each student's characteristics and needs not only conforms to the guiding ideology of South Korea' education policy but can also better meet the learning needs of different students and promote overall development.

It is suggested that a regional shared evaluation case library should be developed later to promote thetration of evaluation reform from "elite education" to ordinary classrooms. Establishing a shared platform containing a wealth of evaluation cases can benefit more teachers and students, thus promoting the of evaluation reform on a larger scale. This not only improves the quality of teaching in ordinary classrooms but also allows more students to enjoy high-quality educational resources and achieve educational equity

References

[1] Ministry of Education of Korea. Fifth Comprehensive Plan for the Promotion of Talent Education (2023-2027) [R. Seoul: Ministry of Education, 2023.

[2] OECD. Report on the Evaluation of Computer Literacy Education in Korea [R]. Paris:ECD Publishing, 2023.

[3] Ministry of Education of Korea. Mathematics Classroom Observation Guide (2025 Trial Edition) [Z] Seoul: National Institute for Educational Research, 2025.

[4] Park, Y. S. Development of Mathematical Evaluation Tools Based on Cognitive Diagnostic [J]. Korean Journal of Mathematics Education, 2025, 38(2): 112-129.

[5] O. Ethical Challenges and Responses in Classroom Assessment Technologies [R]. Paris: OECD Publishing, 2025