Tanuj2005/KG-MCP-SERVER

Knowledge Graph MCP Server

A Model Context Protocol (MCP) server providing seamless access to multiple freely available knowledge graphs including Wikidata, DBpedia, and ConceptNet.

Overview

This server enables AI assistants and applications to query structured knowledge from some of the world's largest public knowledge bases. It provides a unified interface for executing SPARQL queries and searching entities across different knowledge graph platforms.

Features

• Multi-Graph Support: Query Wikidata, DBpedia, and ConceptNet through a single interface
• SPARQL Queries: Execute complex SPARQL queries against Wikidata and DBpedia endpoints
• ConceptNet Integration: Explore semantic relationships and common-sense knowledge
• Async Operations: Built with async/await for efficient concurrent requests
• Error Handling: Comprehensive error handling with timeout protection and detailed error messages
• MCP Protocol: Fully compatible with the Model Context Protocol for AI assistant integration

Supported Knowledge Graphs

Installation

Prerequisites

• Python 3.11 or higher
• uv (recommended) or pip

Setup

1. Clone the repository:

git clone <your-repo-url>
cd KG_MCP_SERVER

2. Install dependencies:

# Using uv (recommended)
uv sync

# Or using pip
pip install -e .

Usage

Running the Server

python main.py

The server will start and expose MCP tools for querying knowledge graphs.

Available Tools

1. list_knowledge_graphs()

List all available knowledge graphs and their capabilities.

Returns:

{
  "available_graphs": {...},
  "total_count": 3
}

2. query_wikidata_sparql(sparql_query: str)

Execute SPARQL queries against Wikidata.

Example:

query = """
SELECT ?item ?itemLabel WHERE {
  ?item wdt:P31 wd:Q5.
  SERVICE wikibase:label { bd:serviceParam wikibase:language "en". }
}
LIMIT 10
"""
result = await query_wikidata_sparql(query)

3. query_dbpedia_sparql(sparql_query: str)

Execute SPARQL queries against DBpedia.

Example:

query = """
SELECT ?person ?name WHERE {
  ?person a dbo:Person .
  ?person foaf:name ?name .
}
LIMIT 10
"""
result = await query_dbpedia_sparql(query)

4. query_conceptnet(concept: str, limit: int = 10)

Query ConceptNet for relationships about a concept.

Example:

result = await query_conceptnet("dog", limit=10)
# Returns relationships like IsA(animal), UsedFor(companionship)

Configuration

Knowledge graph endpoints are configured in . You can add or modify endpoints by editing the KNOWLEDGE_GRAPHS dictionary.

Example Queries

Find All Nobel Prize Winners (Wikidata)

SELECT ?person ?personLabel ?prizeLabel WHERE {
  ?person wdt:P166 ?prize.
  ?prize wdt:P31 wd:Q7191.
  SERVICE wikibase:label { bd:serviceParam wikibase:language "en". }
}
LIMIT 20

Get Information About a City (DBpedia)

SELECT ?city ?population ?country WHERE {
  ?city a dbo:City ;
        dbo:population ?population ;
        dbo:country ?country .
  FILTER(?population > 1000000)
}
LIMIT 10

Explore Concept Relationships (ConceptNet)

# Query for "coffee"
result = await query_conceptnet("coffee", limit=15)
# Returns edges like: UsedFor(staying awake), AtLocation(cafe), etc.

Error Handling

All tools return structured responses with success indicators:

{
  "success": True,
  "results": {...},
  "query": "..."
}

Or on error:

{
  "success": False,
  "error": "Error type",
  "message": "Detailed error message"
}

Resources

• Wikidata Query Service
• DBpedia SPARQL Endpoint
• ConceptNet API Documentation
• Model Context Protocol
• SPARQL Tutorial