Supplements — EHDS Portal

Supplementary materials for the ISWC 2025 Resource Track paper. All files are part of the ehds-linked-data-portal repository and released under Creative Commons Attribution 4.0 International licence.

Documentation

📄

Dataset Description

30 clinical cohorts, ODRL policies, HealthDCAT-AP catalogue schema, named graph structure, and infrastructure overview

dataset_description.pdf

—

Download

📄

Annotation Guide

Atomic fact extraction protocol, completeness and hallucination scoring rules, and five worked examples with highlighted responses

annotation_guide.pdf

—

Download

Evaluation harness

🐍

evaluation_50_queries.py

Runs all 50 benchmark queries under baseline, RAG, and MCP conditions; writes per-query JSON results

eval/evaluation_50_queries.py

—

Download

🐍

benchmark.py

50 benchmark queries with embedded SPARQL ground-truth derivations; authoritative source for all ground truth values

eval/benchmark.py

—

Download

📊

benchmark.csv

Flat CSV version of the benchmark ground truth, one row per query

eval/benchmark.csv

—

Download

Data pipeline

🐍

fhir_to_rdf.py

Converts Synthea FHIR R4 JSON bundles to RDF Turtle, preserving Patient, Condition, Observation, MedicationRequest, and Encounter resources

fhir_to_rdf.py

—

Download

MCP connector

🐍

server.py

MCP server exposing seven typed tools over SSE; compatible with Claude, mcphost, DeepSeek, and the Python MCP SDK

mcp/server.py

—

Download