Question Database
Last updated
Last updated
VEGA AI makes it simple to create, organize, and manage your own question database. Whether you're uploading questions from your existing bank or generating new ones with AI, the platform provides a user-friendly interface and detailed options to ensure every question fits seamlessly into your course structure.
When you navigate to the Question Database tab, you’ll see a clean, organized layout designed for easy access and use. Here’s what you’ll find on the screen:
1. Search and Filter Options
At the top of the interface, there are several filters to help you quickly find specific questions.
Search Bar: Type keywords to find questions based on content or topics.
Filters: Refine your search using options like:
Difficulty: Find Easy, Medium, or Hard questions.
Status: Filter questions that are Verified, Unverified, or AI-Generated.
Subject: Narrow down by specific subjects like Algebra or Geometry.
Topic: Focus on specific topics like Quadratic Equations or Probability.
Question Type: Choose between question types like MCQs, Open-Ended, or Passages.
Use Case: Filter questions based on where they’re used (Practice Tests, Play Tests, etc.).
These options help you quickly locate the exact questions you need, saving time when managing larger databases.
2. Add Questions Button
On the right side of the interface, you’ll see a prominent Add Question button. This is where you start if you’re creating new questions.
3. Generate Questions
Next to the Add Question button, there’s a Generate Questions option. This feature uses AI to create fresh, unique questions based on your specified topic, subject, and difficulty level.
4. Question Dashboard
This section displays a table with all your saved questions. Each question has the following columns:
Question ID: A unique identifier for each question.
Question Text: The main question content.
Status: Indicates whether the question is verified or unverified.
Difficulty: Shows whether the question is Easy, Medium, or Hard.
Subject & Topic Name: Displays the subject and topic the question belongs to.
Actions: Options to View, Verify, Archive, or Generate Similar Questions.
5. Status Indicators
Each question in the table has a status:
Verified: Indicates the question has been reviewed and approved.
Unverified: Questions that need to be reviewed before use.
AI-Generated: Questions created by the AI that require verification.
VEGA AI offers two ways to add questions to your database:
You can upload questions from your existing question bank or type them in manually.
Extract from PDF or Images
Click Add Question and select Extract from PDF/Image.
Upload a high-quality file containing your questions. Ensure each file includes only one clear question per image or page.
Once uploaded, VEGA AI will extract the question text, answer options, and other details automatically.
Try to Upload a pdf with less than 2-3 pages. If you have a pdf bigger than that either split it or take screenshot of the key areas.
Important Tips:
Ensure the document is legible.
Avoid uploading blurry or cluttered images.
VEGA AI is smart, but it works best when you give it the right type of question. If you choose the wrong question type, the AI might get confused or give wrong answers.
That’s why choosing the correct question type is one of the most important steps when you upload questions.
Selecting the appropriate question type helps our AI:
Understand the context
Apply the correct evaluation model
Generate accurate answers or feedback
Avoid misclassification or content rejection
If you mark it as the wrong type, it might:
Give the wrong answer
Miss the point of the question
Not work at all
So before uploading, make sure you pick the right type for each question.
Definition: Includes a passage body followed by one or more questions. There could be just one passage and multiple questions based on the context given in the passage.
Best for: Reading comprehension, evidence-based reasoning, or contextual analysis.
Why it matters: Our AI uses the passage to extract context and meaning. If the passage is missing or mismatched, the system may fail to extract or evaluate the question.
✅ Passage question type example:
Definition: A standalone question with 2 or more choices, only one of which is correct.
Best for: Concept-checking in math, science, grammar, or factual recall.
Why it matters: MCQs should be fully self-contained. If your question needs a passage, do not use MCQ — it will result in misinterpretation.
Avoid using this type for passage-based or dependent questions.
✅ MCQ question type example:
Definition: A question where the student needs to calculate and write a number as the answer. (not multiple choice). So in simple words, the correct answer would always be a number.
In Numeric Question Type, there can be more than one correct answer, as long as they all have the same numerical value.
These are the same type of questions known as Student-Produced Responses (SPRs) in the SAT exam.
Best for: Math problems, physics calculations, or any computation-based tasks.
Why it matters: VEGA AI expects a clear numeric structure. Avoid combining this with other types like MCQ or passage-based questions.
✅ Example:
Definition: Questions requiring a written response, typically 2–5 sentences.
Use this when you want the student to:
Explain something
Share their opinion
Write a reason or analysis
VEGA AI can use rubrics to evaluate open-ended or essay-type questions.
This means the AI doesn’t just check for the right answer — it also looks at how well the response is written. It can assess:
Clarity of explanation
Structure and organization
Grammar and language use
How well the student supports their answer
So when students write short or long answers, VEGA AI can give feedback based on real grading criteria, just like a human teacher would.
✅ Example: “Explain why El Paso became a hub for Spanish-language newspapers in the late 1800s.”
To ensure smooth processing and high-quality output:
Upload Quality
Ensure all files (PDFs, images) are high-resolution and easy to read.
Type Selection
Choose the correct question type for each upload. Do not mix types.
One at a Time
Upload only 1 question per extraction to maximize clarity and accuracy.
Processing Time
Most extractions take 1–2 minutes to process.
Verify Output
Double-check the extracted question text, especially LaTeX formatting, before saving.
Passage
Reading comprehension
MCQ
Factual/standalone logic
Numeric
Calculation-based problems
Open-Ended
Reasoning, writing tasks
Even though VEGA AI does a great job extracting questions, some formatting issues and small errors can happen especially when working with PDFs or scanned documents.
Here’s a complete list of the most common mistakes, how often they happen, and what you can do to fix or avoid them.
Issue
What Happens
How Often?
How to Fix
Things to Watch Out For
1. Italics Not Detected
Italic words don’t appear italic after upload.
~95%
Use *this is italic* in the editor.
Italics won’t show up from PDFs. Add formatting yourself using LaTeX or manually.
2. Bold Not Detected
Bold text becomes regular text.
~95%
Use **this is bold** in the editor.
Bold words may lose meaning. Add bold formatting manually.
3. Underline Not Detected
Underlined words are not preserved.
~95%
Use \underline{your text} with LaTeX or in Mathpix.
Important underlined parts might go missing. Always re-check and add them back.
4. Fill in the Blanks Misformatted
Blanks appear as {qquad} instead of empty lines.
~95%
Replace {qquad} with __________.
{qquad} is hard to read. Always replace with blank lines manually.
Issue
What Happens
How Often?
How to Fix
Things to Watch Out For
5. Wrong Option Marked as Correct
The AI may mark the wrong option as correct.
~30–40%
Double-check with your original answer key.
Never assume the AI-selected answer is right. Always verify.
6. Incorrect Solution Generated
The explanation doesn’t match the correct answer.
~40–50%
Review and rewrite the explanation if needed.
AI explains the answer you select as correct — if that’s wrong, the explanation will also be wrong.
Issue
What Happens
How Often?
How to Fix
Things to Watch Out For
7. Wrong Subject or Topic
The question is placed in the wrong subject or has no topic.
~40%
Choose the correct subject/topic manually in the editor.
Especially important for mixed-topic questions. Check and update tags carefully.
Issue
What Happens
How Often?
How to Fix
Things to Watch Out For
8. Image Not Extracted Properly
Images might not appear or may be blurry.
~0–5%
Upload the image again directly while editing the question.
Always preview the question and make sure images are showing correctly.
9. Options Missing After Extraction
One or more choices don’t show up.
~2–3%
Use the "Edit Question" tool to add missing options.
Do a quick check of every question’s options before saving.
Select Enter Manually when adding a question.
You can chose the type of questions you want to enter from multiple options like MCQs, Passage based, Numerical Response Type etc.
Fill in the following details:
Question Text: Type your question here. Use plain text or mathematical formulas as needed.
Answer Options: Provide up to six possible answers. Mark the correct one(s).
Solution: Write a detailed explanation or create the solution using our AI to help students understand the correct answer.
This method gives you full control over question details.
This feature is ideal for quickly creating new questions.
Click the Generate Questions button.
Select the subject, topic, and difficulty level.
VEGA AI will create a set of questions complete with options, solutions, and explanations.
This is an experimental tool. While we've tried to be as precise as possible, especially in STEM subjects, it's important to be aware of its current limitations:
1. Vision Capabilities: Currently, our model does not support visual understanding. It cannot interpret images within questions nor create new images. We plan to include image understanding in the near future.
2. Accuracy in Mathematics: The model may occasionally make errors, particularly in complex areas such as Geometry and multi-step Mathematical reasoning.
3. Difficulty in Generating Complex Questions: When generating questions of higher difficulty, the model might experience delays or become unresponsive.
4. Factually Incorrect Outputs: At times, the model might generate statements that sound correct but are factually inaccurate. We are continuously working to reduce these occurrences.
5. Grammatical Errors: The model may make grammatical errors during generation. We encourage users to review and correct any mistakes.
Once a question is added or generated, you can review and make changes before saving it to the database. Here's how the customization process works:
Preview the Question
After entering or extracting a question, the system will display a Preview window.
You’ll see the question text, answer options, and solution laid out exactly as students will see it.
This step allows you to verify that everything is accurate and properly formatted.
Edit Question Details
If needed, click the Edit button to make changes. You can:
Modify the question text to make it clearer.
Adjust the answer options or change the correct answer.
Update the solution to provide more detailed or step-by-step explanations.
Add Tags and Classify the Question
Tags help organize your database and make questions easier to find.
Select tags like Pattern Recognition, Integer Identification, or Range Analysis to describe the skills or concepts the question covers.
Classify the question by:
Subject: Choose the relevant subject like Algebra, Geometry, or Data Analysis.
Topic Name: Assign it to a specific topic like Quadratic Equations or Mean, Median, Mode.
Difficulty: Use the slider to assign a difficulty score from 0-100 or select Easy, Medium, or Hard.
Save the Question
Once satisfied, click Preview & Save. The question will now appear in your dashboard.
After building your database, use the interface to keep it organized and up to date:
Search, Filter, and Sort
Use the Search Bar and filters to locate specific questions quickly.
Sort questions by difficulty, status, or subject to manage them efficiently.
Verify and Update Questions
Unverified questions need to be reviewed before they can be included in tests.
Click Verify in the Actions column to mark a question as approved.
If you spot errors, click Edit to update the question or Archive to remove it from active use.
Track Question Usage
Each question shows where it’s being used, such as in Practice Tests or Full-Length Tests. This ensures you know how each question contributes to your course.
Flexibility: Add questions manually, upload existing ones, or let AI generate them.
Customization: Every question can be tailored to match your exact requirements.
Organization: Use tags, filters, and classifications to keep your database structured.
Efficiency: Generate multiple similar questions to save time while maintaining variety.
