Kantor RAG

A RAG-powered research system for exploring the complete works of J.R. Kantor, the founder of interbehavioral psychology.

Overview

Kantor RAG is a Streamlit web application that lets you semantically search through J.R. Kantor's entire body of work — 19 books, 91 articles, and 21 reviews spanning from 1915 to 1984. It uses vector embeddings and retrieval-augmented generation (RAG) to answer research questions grounded in the source texts, with citations.

Features

• Semantic search across Kantor's complete works using sentence embeddings
• RAG-powered answers via Groq's Llama 3.3 70B model, grounded in actual source passages
• Source citations — every answer references the specific book, article, or review it came from
• Filters by document type (Books, Articles, Reviews) and by specific title
• Source diversification — results are drawn from multiple documents, not just one
• Downloadable results — export your query, answer, and sources as a text file
• Clean academic UI styled around Kantor's work

Tech Stack

Component	Technology
Frontend	Streamlit
Vector database	Pinecone
Embeddings	all-MiniLM-L6-v2 (sentence-transformers)
LLM	Groq API — Llama 3.3 70B
Language	Python

Getting Started

Prerequisites

• Python 3.9+
• A Pinecone account and index named kantor-rag
• A Groq API key

Installation

git clone https://github.com/jcampod/kantor-rag.git
cd kantor-rag
pip install -r requirements.txt

Configuration

Copy the secrets template and fill in your API keys:

cp secrets_template.toml .streamlit/secrets.toml

Edit .streamlit/secrets.toml:

PINECONE_API_KEY = "your-pinecone-api-key"
GROQ_API_KEY = "your-groq-api-key"

Run

streamlit run app.py

Corpus

Type	Count
Books	19
Articles	91
Reviews	21
Total	131

Works span from 1915 to 1984 and cover Kantor's core contributions to interbehavioral psychology, psychological linguistics, the philosophy of science, and social psychology.

License

This project is for research and educational purposes.