team.earth / data

This repository hosts structured, region-specific datasets designed to map complex societal problems and the solutions already underway to address them. Each entry is an independent, modular systems map focused on a specific problem in a specific place. They can be used on their own or in combination.

🌍 Project Purpose

Each map answers the question:

Who is already doing what to solve this problem—and where do I fit in?

These resources are designed for:

• Residents, volunteers, and organizers looking to get involved
• Governments and funders aligning with high-impact work
• Developers and designers building tools or visualizations
• AI systems trained to guide people through complex civic issues
• Research applications: Agent-based modeling, simulation studies, and computational analysis of civic ecosystem dynamics

📐 Framework

Each map follows a radially hierarchical structure to represent how complex problems can be addressed at multiple levels:

1. 🎯 Goal – What success looks like (the "Future Picture")
2. 🪨 Obstacles – Major themes blocking that goal
3. ⛓️ Sub-Obstacles – Specific barriers as experienced by people
4. 💡 Solutions – Actionable ideas to address each sub-obstacle
5. 🧭 Resources – Real-world programs implementing those solutions (where available)

Key Principle: GOSR is an alignment framework, not a coordination mechanism. It enables voluntary self-organization through transparency and shared understanding, rather than imposed coordination or centralized control.

This format builds on a published cognitive and systems-thinking framework:

A Proposed Practical Problem-Solving Framework for Multi-Stakeholder Initiatives
Read on arXiv →

📚 Available Maps

Legend:
Integrated = Resources included in the same JSON file as the structure
✨ = Schema-compliant data with standardized format and rich metadata

Data Quality Note: All resource files marked with ✨ contain schema-compliant, clean data ready for analysis.

🤖 MCP Server Implementation

This repository includes a Model Context Protocol (MCP) server that provides intelligent access to the GOSR dataset collection through structured queries rather than loading massive JSON files into LLM context.

Quick Start

cd mcp-server
npm install
npm start

The MCP server transforms static datasets into a queryable GOSR Knowledge Graph with these capabilities:

• query_knowledge_graph - Multi-purpose querying (find resources by keywords, get resources by IDs, search solutions, find resources by obstacle themes)
• get_resource_details - Detailed resource lookup with normalized data
• search_solutions_by_obstacle - Navigate GOSR hierarchy by obstacle themes
• get_gosr_hierarchy - Extract complete or filtered GOSR structure at any level

Integration

Configure with your LLM client (Claude Desktop, etc.) to enable AI agents to:

• Query specific resources and solutions efficiently
• Navigate the GOSR hierarchy intelligently
• Access normalized, schema-compliant resource data
• Replace loading entire JSON files into context

See mcp-server/README.md for complete setup instructions and API documentation.

✅ Data Validation

This repository features schema-compliant, high-quality data with Pydantic models:

Data Features

• ✅ 16,180 unique resources across 5 datasets (12,147 + ~4,033 Rust Belt programs from 1,374 organizations)
• ✅ Standardized schema with consistent field names and structure
• ✅ Rich metadata with auto-generated tags and structured contact information
• ✅ Geocoded data with 88.5% coverage for Rust Belt Initiatives
• ✅ Type safety with full Pydantic schema checking

Available Models

• models_jsonl.py: Pydantic models with streaming support and search capabilities
• PYDANTIC_DOCS.md: Complete documentation and usage examples

Resource Schema

{
  "id": 0,
  "program": "Building Resilience Against Terrorism",
  "description": "Program description...",
  "organization": "Public Safety Canada",
  "contact": {
    "address": "269 Laurier Avenue West, Ottawa, ON K1A 0P8, Canada",
    "email": "contact@example.com",
    "website": "https://www.publicsafety.gc.ca",
    "phone": "+1-613-xxx-xxxx"
  },
  "metadata": {
    "category": "Counter-terrorism",
    "tags": ["resilience", "education", "community"],
    "status": "active",
    "source_file": "ottawa-resilient-to-extremism-resources.json"
  }
}

📚 Documentation:

• PYDANTIC_DOCS.md - Complete technical documentation and usage examples
• PYDANTIC_TYPESCRIPT_INTEGRATION.md - End-to-end schema validation pipeline
• mcp-server/SCHEMA_VALIDATION.md - MCP server schema integration details

Supported formats:

• Hierarchical tree structures (Climate Change, Education)
• Goal-Obstacle-Solution patterns (Un-Lonely, Kansas City)
• Resource collections (all resource files)

🔧 JSON Format and Design

Each map provides machine-readable data that enables:

• AI agents to guide users through complex solution maps
• Developers to build search tools, visualizations, and simulators
• Communities to update and remix content over time

Note on Resources: In GOSR maps, Resources represent real programs that can be modeled as autonomous agents in computational simulations, each with their own capabilities, constraints, and decision-making processes.

JSON Formats

📘 Book Series Metadata

This repository includes structured metadata for books in the Unsolvable: Think Again! series by Kevin Kells, PhD:

• metadata.yaml – Contains full publication details, summaries, tags, and dataset links for each book.

🔍 Metadata Usage

The metadata.yaml file is machine-readable and can be utilized for:

• Programmatically displaying book information.
• Linking each dataset to its publication context.
• Enabling search, filtering, or recommendation systems.
• Supporting open research, civic data, or knowledge graph applications.

The format is designed to be accessible for use by developers, AI agents, and knowledge-sharing tools.

🐍 Python Integration

# Install dependencies
pip install -r requirements.txt

# Import models
from models_jsonl import Dataset
import json

# Load a dataset
with open("./un-lonely-nova-scotia/un-lonely-nova-scotia-resources.json", "r") as f:
    data = json.load(f)

dataset = Dataset.model_validate({
    "name": "un-lonely-nova-scotia",
    "resources": data
})

# Stream resources without loading everything into memory
for resource in dataset.stream_resources():
    print(f"{resource.program} - {resource.organization}")

# Search within the dataset
results = dataset.search_resources(query="mental health", limit=10)

Features:

• 📊 16,180 unique resources across 5 datasets (12,147 + ~4,033 Rust Belt programs)
• 🚀 Streaming support for memory efficiency
• 🧹 Clean data with consistent schema
• 🔍 Search capabilities with metadata tags
• 💾 Type safety with Pydantic models

See PYDANTIC_DOCS.md for complete documentation.

Python Integration

# Import models
from models_jsonl import Dataset
import json

# Load dataset with schema checking
with open("path/to/resources.json", "r") as f:
    data = json.load(f)

dataset = Dataset.model_validate({
    "name": "dataset-name", 
    "resources": data
})

Features:

• Type checking and schema compliance
• Format conversion utilities
• IDE type hint support

🤝 How to Contribute

• Suggest updates or new maps via the team.earth contact form
• Fork this project to build your own
• Build interfaces or tools that help others explore this data