The University Admission Chatbot is an intelligent conversational assistant designed to automate and simplify access to university-related information. Using advanced Natural Language Processing (NLP) and Retrieval-Augmented Generation (RAG) techniques, the chatbot provides instant, accurate, and human-like responses to student queries about admissions, fees, scholarships, and courses.
🔗 Deployed Link:
https://university-admission-chatbot-fronte.vercel.app/
| Name | Student ID |
|---|---|
| Jenil Soni | 23BCE119 |
| Rishi Kukadiya | 23BCE156 |
| Karmit Langhnoda | 23BCE157 |
| Mahin Mehta | 23BCE161 |
| Anas Multani | 23BCE188 |
- Natural Language Understanding: Interprets user queries and intent using advanced NLP models.
- Slot Extraction: Extracts key details (e.g., course name, location, or fee type) from user input.
- Intent Classification: Determines the user’s purpose (e.g., admission process, fee details, scholarships).
- Document Retrieval (RAG): Searches relevant university PDFs or scanned documents using TF-IDF and cosine similarity.
- Context-Aware Conversations: Maintains user context and conversation flow using MongoDB.
- Human-like Responses: Generates answers via Google Gemini 2.5 Flash, ensuring factual and conversational accuracy.
- Scalable Design: Modular architecture for easy integration with other university services.
- Multilingual Support: Easily extendable to handle queries in multiple languages.
- User Input — The student types a question into the chat interface (React frontend).
- Backend Processing — Backend creates/updates user session and forwards query to AI agent.
- Intent & Slot Extraction
- Intent Classifier: A HuggingFace-based Transformer model identifies user intent.
- Slot Extractor: Uses Google Gemini 2.5 Flash (via LangChain) to parse structured data (JSON format).
- Retrieval-Augmented Generation (RAG):
- Retrieves top relevant document chunks via TF-IDF + cosine similarity.
- Combines retrieved content with user context to craft a factual LLM prompt.
- Response Generation: Gemini 2.5 Flash produces a domain-specific, human-like response.
- Database Logging: Session data and extracted slots are stored in MongoDB.
- Final Output: The chatbot delivers a precise and contextual response to the user.
| Component | Model / Framework | Function |
|---|---|---|
| Slot Extraction | Google Gemini 2.5 Flash (via LangChain) | Converts natural text into structured JSON data. |
| Intent Classification | HuggingFace Transformers | Classifies user intent (e.g., fees_info, admission_process). |
| Document Retrieval | TF-IDF + Cosine Similarity | Finds the most relevant university documents. |
| OCR Processing | PyPDF2 + Tesseract OCR | Extracts text from scanned or image-based PDFs. |
| Database | MongoDB | Stores session context and slot data persistently. |
| Frontend | React.js | Provides an interactive chat interface. |
| Backend | Python (Flask / FastAPI) | Handles NLP processing and API integration. |
| Layer | Technologies |
|---|---|
| Frontend | React.js, HTML, CSS |
| Backend | Python, Flask / FastAPI |
| NLP / AI | LangChain, HuggingFace Transformers, Gemini 2.5 Flash |
| Retrieval | TF-IDF, Cosine Similarity |
| OCR & Data | PyPDF2, Tesseract OCR |
| Database | MongoDB |
| Version Control | Git & GitHub |
- High Accuracy: RAG ensures the chatbot delivers factual, document-grounded answers.
- Persistent Context: MongoDB enables continuous conversation flow and session management.
- Scalable & Modular: Components can be independently improved or replaced.
- Real-world Application: Demonstrates the power of NLP and information retrieval for academic institutions.
- Voice-based query support (speech-to-text & text-to-speech integration)
- Advanced multilingual support (regional languages)
- Integration with student portals for personalized information
- Enhanced analytics dashboard for administrators
The Gujarat University Admission Chatbot showcases the fusion of NLP, document retrieval, and conversational memory to build intelligent, context-aware systems. It represents a major step toward automating university information systems and improving student engagement through natural, conversational interactions.
For project inquiries or collaborations, contact the team via email or through the university’s innovation cell.