Adaptive and Explainable Quiz Generation

### Description
The current quiz generation system creates short quizzes from user-provided content.
While functional, the generated quizzes are **static** and lack mechanisms for difficulty control, answer explainability, and quality validation.

This issue proposes enhancing the existing quiz pipeline to improve **learning quality, reliability, and adaptability**.

---

### Current Limitations
- No control over question difficulty
- No explanations linked to source content
- Possible duplicate or ambiguous questions
- No use of user performance for quiz adaptation

---

### Proposed Enhancements

#### 1. Difficulty-Controlled Question Generation
- Categorize questions as Easy / Medium / Hard
- Allow users to select preferred difficulty
- Avoid overly trivial or repetitive questions

---

#### 2. Source-Grounded Answer Explanations
- Attach explanations to each question
- Link correct answers to the exact content segment they are derived from
- Ensure all questions are verifiable from the input content

---

#### 3. Adaptive Quiz Flow
- Track user performance (accuracy, response time)
- Dynamically adjust question difficulty and topic focus
- Generate follow-up questions based on weak areas

---

#### 4. Question Quality Control
- Detect and remove duplicate or near-duplicate questions using embedding similarity
- Identify ambiguous questions with multiple valid answers
- Enforce a single-correct-answer constraint

---

### Technical Notes
- Focus on backend logic and NLP pipeline improvements
- Minimal UI changes
- Modular and extensible design for future enhancements

---

### Expected Outcome
- Higher-quality quizzes with controlled difficulty
- Improved learner engagement and trust
- Better educational value through adaptive assessment