Uedu Worksheet
Competence-based worksheets

1. Overview

Uedu Worksheet (Worksheet Workshop) is a worksheet generation and management system centred on the competence-based approach of the 108 Curriculum Guidelines. The system can automatically generate draft worksheets aligned with the five competence levels through AI, from classroom recording transcripts, RAG teaching materials, or text provided by the teacher, and then the teacher can refine them before issuing them to Students as homework.

The design of the Worksheets is based on the competency-based teaching design guidelines of the National Academy for Educational Research (NAER) and the College Entrance Examination Center (CEEC), with emphasis on:

• Integrating knowledge, skills and attitudes: Assessing not only knowledge points, but also whether Students can apply and reflect
• Contextual learning: linking classroom learning with real-life situations
• Emphasise Learning history: guide students to become aware of their own thought processes
• Practical application: enable Students to transfer what they have learned to new contexts

2. Structure of the five levels of literacy

Each worksheet organises questions according to the following five levels, guiding Students from shallow to deep thinking:

3. Four generation sources

Instructors can create Worksheets from four different sources:

4. AI generation mechanism

The process for AI-generated Worksheets is as follows:

1. Instructor selects source, subject, year, difficulty, number of questions
2. The system creates a background generation task and obtains the source text
3. Build a subject-specific System Prompt (including the five-level competency structure, question-type requirements, and assessment rubric format)
4. Call GPT-4.1 (JSON mode) to generate a complete Worksheet
5. Validate the response format and fill in default values (marking mode, scores, etc.)
6. Write to the database, and set the status to "Draft"
7. After reviewing and fine-tuning in the editor, the Instructor can assign it

Generation parameters

Automatic mapping between question type and marking mode

5. Hybrid marking mode

Worksheets use a "mixed marking" design, balancing immediate feedback with Instructor professional judgement:

Multiple-choice questions: automatic real-time marking (auto)

• After students submit, multiple-choice / true-false / fill-in-the-blank questions immediately display correctness and scores
• Compared with the standard answer, 0 or full marks (no partial credit)
• Instructors can set whether the correct answer is shown at the same time

Open-ended question: AI pre-marking + Instructor confirmation (ai_review)

1. After students submit, GPT-4.1-nano is called in the background to generate suggested scores and feedback according to the rubric
2. Students see "submitted, awaiting teacher review"
3. On the teacher side, AI-recommended scores and feedback text can be:
   • One-click adopt: use the AI suggestion directly
   • Confirm after editing: adjust scores or rewrite feedback
   • Batch accept: accept the AI suggestions for all students' open-ended answers at once
4. After the instructor confirms, the feedback is published for students

6. Instructor workflow

1. Generate Worksheets: Choose a source (transcript/textbook/pasted text/blank), set the subject, year group, difficulty and number of questions
2. Edit draft: review AI-generated questions in the Worksheet editor; you can add, delete and revise question text, adjust competency levels, and edit options and marking rubric
3. Distribute: set the opening time, closing time and maximum number of attempts, then distribute to the whole class of Students
4. Monitoring: View submission rates and question-by-question accuracy statistics on the marking dashboard
5. Marking: review AI-premarked open-ended questions (mark one by one or accept in batches), and write overall feedback
6. Export: export PDF for printing or archiving

7. Student workflow

Students can enter the Worksheets from two entry points:

• Student Console: the course card shows a “To be studied Worksheets” badge; click to expand the list and go straight into answering
• Course classroom page: right-hand "Worksheets" tab, listing all worksheets and their status

Response process:

1. Tap "Start answering" and the system records the start time
2. Answer each question in order; drafts are saved automatically every 30 seconds
3. The progress bar at the top shows response progress (N/M questions answered)
4. After confirming and submitting:
   • Multiple-choice questions: immediate display of correctness and score
   • Open-ended question: display 'Submitted, awaiting Instructor marking'
5. After the instructor has finished marking, students can view full feedback and scores

8. PDF export

Worksheets can be exported as A4-sized PDF files, suitable for printing and giving to Students for handwritten responses. The PDF includes:

• Worksheet title, subject, year, estimated time
• Name / seat number / date fields
• Questions grouped by literacy level
• Answer selection area for multiple-choice questions
• Blank answer area for open-ended questions (short answer 1.5cm / essay 4cm)
• Uedu brand footer

PDFs are generated server-side with ReportLab, with support for Chinese fonts (Noto Sans TC).

9. Research applications

The data generated by Uedu Worksheet can support the following directions for educational research:

Research areas

• Differences in performance across literacy levels: compare the distribution of scores across the five literacy levels to identify weaknesses in the class at the "reflection" or "transfer" levels
• AI-generated quality assessment: comparing learning Worksheets generated by AI with those designed manually by teachers in terms of student learning outcomes
• Hybrid marking consistency: Analyse the consistency between AI-suggested scores and the instructor's final scores (Cohen's Kappa) to assess the credibility of AI marking
• Class recording → Worksheet automation workflow effect: quantify the time it takes an instructor from recording to distributing Worksheets, and the efficiency gains compared with traditional manual design
• Interdisciplinary literacy assessment: use discipline-specific prompts to compare design patterns for literacy questions across subjects

Exportable data