No idea where to start with final exam review? Use a personal knowledge base to turn courseware into self-test questions
The least efficient thing you can do when reviewing for finals is to flip through the courseware from beginning to end. While flipping through it, you feel like you understand everything; once you close it, you can't recall a thing.
Later, I changed my approach:don't retell it—just make questions.
Step 1: First, gather a semester's materials in one place
Put the courseware, handouts, assignments, lab reports, and the exam points the teacher emphasized in class all in one place. In the past, they were scattered across four folders and three group chats, and finding materials during review took more time than the review itself.
For this step, I used MuchSci's personal knowledge base. The official line is “let existing knowledge support new research”—as I understand it, it isn't a cloud drive; the key is that you can ask questions.
Step 2: Categorize materials by “how the exam will ask about them”
I don't categorize by chapter; I categorize by question type:
· Concept questions: How is this term defined, and how does it differ from similar terms?
· Calculation questions: formulas, conditions, typical steps
· Short-answer questions: why do it this way, and what are the prerequisites?
· Application questions: given a scenario, which method set do you use?
After categorizing, I found that most of the courseware fell into the “concept” and “short-answer” categories, while the calculation part was actually concentrated in just three lessons.
Step 3: Make a batch of self-test questions for each category
Create questions from the materials—one question per knowledge point—and the answer must be findable in my own materials.
I set two rules for myself when writing questions:
1. The standard for an answer is being able to restate it in one sentence, not just memorize it.
2. Every question must cite its source (which lecture, which page). If I can't find a source for a question, that means I may have made up that knowledge point myself.
This step took me the most time, and it also paid off the most—being able to write questions means I've truly absorbed the material. The knowledge points I can't write questions for are exactly the parts I haven't understood.
Step 4: Go through it in Q&A format
After writing the questions, I treat it as an open-book self-test: answer first, then go back to the materials to check.
During the checking process, I find two types of problems: the ones I answered incorrectly (go back and review that lecture), and the ones where “I answered correctly but can't explain why” (these are the most dangerous—if the exam asks it a different way, I fall apart).
Step 5: Keep a “two hours before the exam” checklist
By the end of review, I keep only one page:
· Five concepts that are easiest to confuse (write them in pairs)
· Three formulas that are certain to be tested (including conditions)
· Two things I still haven't figured out (ask classmates the day before the exam)
This page is a distillation of the first four steps. In the two hours before the exam, I only flip through this page.
A little reflection
Turning courseware into self-test questions takes more work than “reading it again,” but the review payoff is much higher. Because “reading” is passive, while “writing questions” forces me to judge what the key points are and what I don't understand.
Now I start doing this in the middle of every semester, so I'm not scrambling at finals.
I've been adding to MuchSci's personal knowledge base one semester at a time. By the third semester, the self-test questions from the first three courses are still there, and they're faster to browse than hunting for old courseware. For an AI assistant for college students, what I need isn't for it to summarize courseware for me; it's to help me organize materials into a format I can repeatedly self-test with.