How rishabhguptax/campusquery-rag works
CampusQuery-RAG answers student questions about college information using retrieved document context and returns source/page references. The main workflow is Streamlit chat → RAG orchestration → semantic retrieval from ChromaDB → grounded Groq generation → answer and citations. A separate PDF-ingestion path extracts page text, chunks it, embeds it, and populates the vector store. README also documents a website-scraping ingestion path in ingest_data.py. The sampled code supports the PDF path and query flow; the documented API-key setup and OCR/evaluation/deployment plans are not separate implemented runtime components.
Main entry points: rag_pipeline.py RAG orchestration · app.py Streamlit chat · pdf_ingest.py Text chunking · llm.py Grounded answer generation
Components 16
Student interface
- Streamlit chat
app.py - Answer with sources
app.py
Question answering
- RAG orchestration
rag_pipeline.py - Semantic retrieval
retriever.py - Context and citations
rag_pipeline.py - Grounded answer generation
llm.py
Knowledge ingestion
- College PDFs
- PDF page extraction
pdf_ingest.py - Text chunking
pdf_ingest.py - College website crawl
ingest_data.py - HTML text cleaning
ingest_data.py - Web page chunking
ingest_data.py
Models and storage
- Embedding model
pdf_ingest.py - Chroma vector store
pdf_ingest.py - Groq service
Other components
- Student
Connections 17
- Student → Streamlit chat: asks
README.md - Streamlit chat → RAG orchestration: submits question
app.py - RAG orchestration → Semantic retrieval: searches
rag_pipeline.py - Semantic retrieval → Chroma vector store: queries
retriever.py - RAG orchestration → Context and citations: builds
rag_pipeline.py - RAG orchestration → Grounded answer generation: requests answer
rag_pipeline.py - Grounded answer generation → Groq service: calls
llm.py - Context and citations → Grounded answer generation: supplies context
rag_pipeline.py - RAG orchestration → Streamlit chat: returns answer and sources
app.py - Streamlit chat → Answer with sources: renders
app.py - College PDFs → PDF page extraction: provides pages
README.md - PDF page extraction → Text chunking: splits extracted text
pdf_ingest.py - Text chunking → Embedding model: embeds chunks
pdf_ingest.py - Text chunking → Chroma vector store: stores chunks
pdf_ingest.py - College website crawl → HTML text cleaning: cleans HTML
ingest_data.py - HTML text cleaning → Web page chunking: provides text
ingest_data.py - Web page chunking → Chroma vector store: stores chunks
ingest_data.py