🚀 FocuPrep AI: Accelerating Competitive Exam Preparation with GenAI

A mini-project on reimagining UPSC preparation using Generative AI

Why this project?

Preparing for competitive exams like UPSC (Union Public Service Commission) is one of the most daunting challenges in India. Every year, millions of aspirants attempt the exam, but only a few succeed.

The difficulty arises from:

• Huge amount of data: Syllabus, NCERT textbooks, current affairs, and optional subject books.
• Strict timelines: Prelims → Mains → Interview, all within a tight schedule.
• Dynamic nature: Staying up to date with current affairs while linking them to static concepts.

The Data Landscape

When preparing for UPSC, here’s what an aspirant needs to master:

• Syllabus (foundation of everything)
• NCERT Textbooks (the “bible” of UPSC prep)
• Newspapers (current affairs that often indirectly link to exam questions)
• Optional Subject Books (in-depth for chosen specialization)

Unlike some exams, UPSC questions are not directly lifted from sources. Instead, they are multi-hop linked:

Example:
If a cyclone hits a coastal city, the exam question may not ask “When did the cyclone occur?”. Instead, it might ask about the culture, geography, or policies relevant to that city—but only if it’s in the syllabus.

So, the key chain is:
Entity/Topic → Syllabus → NCERT or Core References

Too much data overwhelms; the solution is minimalized, relevant information.

What do aspirants really need?

1. Efficient comprehension of massive content in limited time.
2. Smart linking & visualization of data across multiple sources.

Imagine reading the newspaper—
Instead of consuming every article, you only see syllabus-relevant content, and alongside it, you view the connected NCERT references.

Building the GenAI Agent

Step 1: Establish Data Sources

• Syllabus
• NCERT textbooks
• Newspaper / Current Affairs
• Previous Year Question Papers

Step 2: Digitize & Structure the Content

• NCERT Graph
  • Lexical nodes: Chapter → Page → Components
  • Semantic nodes: Entities & topics extracted
• Syllabus Graph
  • Each line as a node → linked entities/topics
• News Graph
  • User-uploaded or live feed → entities/topics extracted

Step 3: Build the Knowledge Graph

Interconnect:
Syllabus Graph ↔ NCERT Graph ↔ Newspaper Graph

Step 4: User Experience

• Left panel: Newspaper (filtered by syllabus relevance)
• Right panel: Linked NCERT references, syllabus line, related topics

This ensures:
✅ Minimal reading, maximal relevance
✅ Faster connections between current affairs & core syllabus

Future Enhancements

1. Adaptive Linking & Ranking
   • Train on previous year papers, textbooks, and news to refine what’s “important.”
   • Reduce overwhelm by filtering noise.
2. Mock Question Generation
   • Use past papers + GenAI to generate probable questions.
   • Provide auto-generated answers for practice.
3. Continuous Learning
   • Model adapts based on user feedback (likes/dislikes on content).
   • Personalized improvement for every aspirant.

Takeaway

This project is more than just an experiment. It’s a vision:

Turning GenAI into a study companion that accelerates competitive exam preparation, saving time and maximizing focus.

If scaled, this could transform how millions prepare for UPSC and other tough exams across the globe.

‍