Skip to main content

​
What is DeenPAL?

DeenPAL is a Retrieval-Augmented Generation (RAG) based chatbot designed to provide accurate and personalized Islamic information to users based on authentic Hadiths from Sahih Al-Bukhari and Sahih Muslim. Unlike traditional chatbots that rely solely on pre-trained knowledge, DeenPAL retrieves relevant Hadiths from a curated database and uses advanced language models to generate contextually appropriate responses with proper citations.

​
Key Features

Personalized Responses

Delivers tailored information based on user queries, ensuring each answer is relevant to your specific question.

Reliable Sources

Utilizes trusted Hadith sources (Sahih Al-Bukhari and Sahih Muslim) to ensure information accuracy and authenticity.

Interactive Interface

Engages users through a conversational platform powered by Streamlit for seamless interaction.

Smart Retrieval

Uses Maximal Marginal Relevance (MMR) to retrieve diverse yet relevant Hadiths, avoiding redundant responses.

​
Technology Stack

DeenPAL is built using modern AI and data processing technologies:
  • LangChain: Framework for building RAG pipelines and managing retrieval-generation workflows
  • ChromaDB: Vector database for efficient semantic search and Hadith retrieval
  • Streamlit: Interactive web interface for user-friendly chat experience
  • DeepSeek: Large language model for generating accurate and contextual responses
  • HuggingFace Embeddings: sentence-transformers/all-MiniLM-L6-v2 for creating semantic embeddings
  • Python 3.10: Core programming language

​
How It Works

DeenPAL follows a structured RAG approach:
1

Data Preparation

Hadith PDFs are loaded from the data/ directory, split into semantic chunks (by chapter and hadith number), and converted into embeddings stored in ChromaDB.
2

Query Processing

When you ask a question, DeenPAL uses semantic search with MMR to retrieve the 4 most relevant and diverse Hadiths from a pool of 10 candidates.
3

Response Generation

The retrieved Hadiths are provided to the language model along with your query to generate a comprehensive answer with proper citations.
4

Citation and Explanation

Each response includes the Hadith text, source details (book number, hadith number, chapter), and a brief explanation before answering your question.

​
Use Cases

DeenPAL is designed for:
  • Students of Islamic Studies: Research authentic Hadith references quickly
  • Scholars and Educators: Find relevant Hadiths with proper citations for teaching
  • General Users: Learn about Islamic teachings from authentic sources
  • Researchers: Explore Hadith collections with semantic search capabilities

​
Benefits

DeenPAL ensures all responses are grounded in authentic Hadith texts, reducing the risk of hallucinations common in traditional LLMs.
  • Accuracy: Responses are based on actual Hadith texts, not generated knowledge
  • Traceability: Every answer includes source citations for verification
  • Diversity: MMR-based retrieval provides varied perspectives from different Hadiths
  • Efficiency: Cached data loading ensures fast response times
  • Transparency: Clear separation between retrieved Hadiths and AI-generated explanations

​
Next Steps

Quickstart

Get DeenPAL up and running in minutes with our step-by-step guide.

Architecture

Learn about the technical architecture and design decisions behind DeenPAL.

Build docs developers (and LLMs) love

Get started for freeTalk to us