TutorialsMeta-Question Method: Escape the Question Sea
Question-sea tactics vs. meta-question induction + systems thinking + 7-step extraction flow + a Gaokao math function case.
Pain Point: The Question Sea Exhausts You
Did 500 questions, still stuck when the exam changes the wrapping? Every question feels new — no knowledge network formed.
Solution: Meta-Question Induction
Group similar questions into one class and extract the "meta-question" — mastering one meta-question = mastering one class of questions.
| Question sea | Meta-question method |
|---|---|
| 500 questions, exhausting | 20 meta-questions, easy |
| Stumped by any change | All variants share one root |
| Can't memorize | Understand the essence, generalize |
Systems Thinking: From "Isolated Drilling" to "Relation Networks"
Systems thinking: treat a question as a system jointly composed of test points, stem conditions, solution methods, variants, and pitfalls. Paying attention to how these elements interact upgrades "memorizing one question" into "mastering one class".
The relational view: knowledge points rarely appear in isolation — concrete questions connect through shared concepts, preconditions, and solution paths. A meta-question is the hub of the knowledge network: once connections are established, new questions can quickly invoke related methods along existing paths.
Activate one meta-question hub → mobilize a whole group of related knowledge (points, conditions, methods, variants, pitfalls) at once.
The 7-Step Extraction Flow
Collect similar questions
Gather 10-50 similar questions from error notebooks, past papers, and workbooks.
Let AI identify common features
Batch-feed the questions to AI and let it identify their common features.
Summarize core test points
Have AI summarize the core test points that recur behind these questions.
Extract the meta-question (minimal form)
Strip away concrete numbers and wrapping; extract the minimal form of the meta-question.
Generate variants (3-5)
Generate difficulty-ascending variants around the meta-question.
Build solution templates
Distill a reusable solution template for each meta-question.
Generate knowledge cards
Turn meta-question → variants → points → pitfalls into knowledge cards for anytime review.
A one-shot prompt:
Extract meta-questions from the following [number] questions:
[paste questions]
Requirements:
1. Identify the common test points of these questions
2. Extract 1-3 meta-questions (minimal form)
3. Generate a solution template for each meta-question
4. Generate 3 variant questions (ascending difficulty)
5. Present in a table: meta-question → variant → point → pitfallDemo Case: Gaokao Math "Functions" Extraction
50 function questions → 6 meta-questions:
| Meta-question | Covered points | Variant directions |
|---|---|---|
| MQ 1: Domain | Fractional / radical / logarithmic | Composite function domains |
| MQ 2: Monotonicity | Definition / derivative method | Parameterized monotonicity |
| MQ 3: Extrema | Quadratic / closed interval | Parameterized case analysis |
| MQ 4: Parity | Definition verification | Abstract function parity |
| MQ 5: Zero points | Zero point theorem | Zero count and parameters |
| MQ 6: Graph transforms | Shift / symmetry / scaling | Function-equation synthesis |
Mastering 6 meta-questions = handling 80% of Gaokao function questions. Next lesson goes hands-on: AI question generation + Socratic questioning to test real understanding.
7 Planning Views + Skill Pack
Gantt, timeline, kanban, heatmap, concentric circle, staircase, progress ring — each with a copy-ready prompt, plus an exam-coach Skill.
AI Question Generation + Socratic Review
5 question types + generation prompt + 10 levels of questioning to test real understanding + 8 review methods and an applicable-exam list.