Paper 5a: Federated Biomedical Knowledge Graphs — GRADES-NDA @ SIGMOD 2026

Title: [Demo] Federated Biomedical Knowledge Graphs: Graph-Native Storage, Planning, and AI-Agent Integration

Venue: GRADES-NDA Workshop at SIGMOD 2026, Bengaluru, India (June 5, 2026)

Status: Submitted

Authors: Madhulatha Mandarapu, Sandeep Kunkunuru

Summary

Demo paper showing three federated biomedical knowledge graphs (Pathways, Clinical Trials, Drug Interactions) queryable through a single Cypher interface. Demonstrates graph-native query planning, cross-KG federation via shared property bridges, and MCP tool integration for AI-agent access.

The full In-Use track companion paper covering the 11-KG federated mega-benchmark is Paper 5b.

Key Results

• 3 KGs federated: 7.9M nodes, 28M edges
• BiomedQA: MCP 98% · Text-to-Cypher 85% · GPT-4o standalone 75%
• Graph-native planner: cost-based selection across multiple candidate plans per query

Manuscript sources, reviewer feedback, and reproduction artifacts live in the private samyama-research repository under papers/paper5-grades-nda/.