safishamsi

Knowledge Graph Based Hybrid RAG System

Name: Knowledge Graph Based Hybrid RAG System
Author: safishamsi

AI community

SOURCE https://github.com/safishamsi/Knowledge-Graph-Based-Hybrid-RAG-System

Description

Knowledge Graph-RAG system using Neo4j, LangChain, LangGraph, Claude-3.5-Sonnet. Hybrid retrieval: SBERT embeddings + graph traversal + BM25. 61K papers, 190K authors from Scopus. 50% NDCG improvement, 57% bias reduction. Includes notebooks, evaluation scripts, benchmark queries. Python/Jupyter.

Installation

This entry records only its repository, not the path inside it, so there is no exact command to give. Open the source below and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

README

Knowledge Graph-Based Hybrid RAG System

![Python](https://img.shields.io/badge/python-3.8+-blue.svg) ![Neo4j](https://img.shields.io/badge/Neo4j-4.0+-green.svg) ![License](https://img.shields.io/badge/license-MIT-blue.svg)

Academic search system combining knowledge graphs with retrieval-augmented generation to eliminate citation bias and hallucinations in research discovery.

Key Features

50% better search relevance (NDCG@10: 0.814) than traditional systems
57.5% reduction in temporal citation bias
67% fewer hallucinations through document grounding
Sub-500ms query response times
Automatic collaboration discovery via community detection

Tech Stack

Knowledge Graph: Neo4j (61,945 papers, 189,972 authors)
Embeddings: SBERT + FAISS indexing
RAG Pipeline: LangChain/LangGraph + Claude-3.5-Sonnet
Data Source: Scopus API integration
Language: Python

Results

82% researcher preference over Google Scholar
64% reduction in literature review time
96% cost reduction vs GPT-4 ($1.02 vs $24/1000 queries)

Project Structure

Knowledge-Graph-Based-Hybrid-RAG-System/ ├── Neo4jKG/ ├── RAG/ ├── embeddings/ ├── scopusscraping/ ├── Dissertation/ ├── Data/ ├── LLMpoweredRAG.py ├── demowithinspector.pdf ├── queries.txt ├── uobkg.png ├── LICENSE └── README.md

Installation & Setup

Prerequisites

Python 3.8+
Neo4j Database
Anthropic API key for Claude-3.5

Installation

git clone https://github.com/safishamsi/Knowledge-Graph-Based-Hybrid-RAG-System.git
cd Knowledge-Graph-Based-Hybrid-RAG-System

# Install dependencies
pip install neo4j sentence-transformers faiss-cpu langchain langchain-anthropic langgraph

# Set environment variables
export ANTHROPIC_API_KEY="your-api-key"

from LLMpoweredRAG import AcademicSearchSystem

# Initialize the system
search = AcademicSearchSystem()

# Search for papers
results = search.query("machine l