An end-to-end NLP system that classifies student queries into learning intents to improve search and content delivery in educational platforms.
🚀 Overview
This project builds a transformer-based text classification system that identifies the intent behind student queries, enabling adaptive routing of educational content.
🎯 Supported Intent Classes:
CONCEPTUAL → Understanding concepts
NAVIGATIONAL → Navigating resources
PROCEDURAL → Step-by-step problem solving
ADVANCED → Deep technical exploration
🧠 Architecture
Data → Preprocessing → Transformer Model → Evaluation → Streamlit UI
⚙️ Tech Stack
Hugging Face Transformers
PyTorch
Scikit-learn
Streamlit
📊 Dataset
Custom dataset (~560 samples)
Manually labeled across 4 intent categories
Balanced class distribution
📈 Model
Base model: DistilBERT
Fine-tuned on domain-specific queries
Evaluation metrics:
Accuracy
Precision / Recall / F1-score
🔬 Key Features
Custom dataset design for educational intent modeling
