Example of Institution Collaboration Network

This repository contains tools for analyzing institutional collaborations using the OpenAlex API. The tools help visualize collaboration networks, trends, and patterns between academic institutions.

📊 Example Visualizations

Global Collaboration Map

Collaboration Trends

Analysis Summary

• Institution Data Fetching: Retrieve and save data about academic institutions worldwide
• Collaboration Analysis: Analyze collaboration patterns between institutions
• Visualizations:
  • Interactive network graphs of institutional collaborations
  • Global collaboration distribution maps
  • Time-series analysis of collaboration trends
  • Summary dashboards
  • Top collaborating institutions networks

First, clone the repository and install the required packages:

git clone https://github.com/Ahanmr/OpenAlex-KG-Analyzer.git
cd OpenAlex-KG-Analyzer
pip install -r requirements.txt

Required packages:

• pandas>=1.5.0
• networkx>=3.0
• plotly>=5.13.0
• requests>=2.28.0
• country_converter>=1.0.0
• numpy>=1.23.0

First, fetch and save institution data from OpenAlex:

python fetch_institutions.py

This creates a directory institution_data_TIMESTAMP containing:

• institutions_full.csv: Complete dataset with all institution details
• institutions_simple.csv: Simplified version with key fields
• top_institutions.md: Markdown file listing top 100 institutions
• summary.txt: Summary statistics about the institutions

To analyze collaborations for a specific institution:

python run_analysis.py

Before running, update these parameters in run_analysis.py:

institution_id = "I2802101240"  # Your target institution ID
email = "[email protected]"  # Your email for API access

This creates a directory results_INSTITUTION-ID_INSTITUTION-NAME_TIMESTAMP containing:

1. Raw Data

   • collaboration_data.csv: Raw collaboration data

2. Interactive Visualizations

   • collaborations_over_time.html: Time series of collaboration counts
   • collaboration_map.html: Global distribution of collaborations
   • collaboration_trends.html: Trends with top collaborating countries
   • institution_network_top20.html: Network of top 20 collaborating institutions
   • institution_network_top50.html: Network of top 50 collaborating institutions
   • collaboration_summary.html: Summary dashboard

3. Network Data

   • collaboration_network.gexf: Network graph file (can be opened with Gephi)

• Shows collaboration relationships between institutions
• Node size indicates collaboration frequency
• Interactive hovering shows collaboration details
• Available in two versions: top 20 and top 50 institutions

• Choropleth map showing global collaboration distribution
• Color intensity indicates collaboration frequency
• Uses country_converter for accurate country naming
• Interactive tooltips with collaboration counts

• Time series visualization of collaboration patterns
• Shows trends for top collaborating countries
• Interactive legend for filtering countries
• Hover information for detailed counts

• Combined view of collaboration metrics
• Top collaborating institutions bar chart
• Yearly collaboration trend line
• Overall collaboration statistics

1. Run fetch_institutions.py
2. Check generated files:
   • institutions_simple.csv
   • top_institutions.md

• Base URL: https://api.openalex.org/institutions
• Search endpoint: https://api.openalex.org/institutions?search=INSTITUTION_NAME

INSTITUTION_IDS = {
    'Harvard University': 'I127803138',
    'Stanford University': 'I127803157',
    'MIT': 'I127803157',
    'University of Oxford': 'I127803157',
    'University of Cambridge': 'I127803160'
}

max_pages = 50  # Number of pages to fetch
per_page = 200  # Results per page

start_year = 2020  # Start year for analysis
end_year = 2023    # End year for analysis
top_n = 20        # Number of top institutions to show

• openalex_id: Unique identifier
• display_name: Institution name
• country_code: ISO country code
• type: Institution type
• works_count: Number of publications
• cited_by_count: Citation count

• year: Publication year
• collaborating_institution: Partner institution name
• country: Collaborator's country code
• work_id: Publication identifier

To construct and analyze co-authorship networks using graph.py:

python graph.py

Before running, update these parameters in the main() function of graph.py:

email = "[email protected]"  # Your email for API access
institution_id = "I97018004"      # Target institution ID
start_year = 2022                 # Start year for analysis
end_year = 2023                   # End year for analysis
max_papers = 500                  # Maximum number of papers to analyze

This creates a directory knowledge_graph_TIMESTAMP containing:

1. Network Data

   • author_nodes.csv: Node-level data with author metrics
   • coauthorship_edges.csv: Edge data with collaboration strengths
   • coauthorship_network.gexf: Network file for Gephi visualization

2. Interactive Visualizations

   • coauthorship_network_top_10.html: Network of top 10 authors
   • coauthorship_network_top_20.html: Network of top 20 authors
   • coauthorship_network_top_50.html: Network of top 50 authors

3. Analysis Files

   • network_metadata.txt: Network statistics and summary
   • graph_generation.log: Processing log with details

The generated network visualizations include:

• Node size based on publication count
• Edge weight based on collaboration frequency
• Interactive hover information showing:
  • Author name and institution
  • Publication count
  • ORCID (when available)
  • Collaboration details
• Color scaling based on publication metrics

• OpenAlex for the comprehensive research data API
• NetworkX for network analysis tools
• country_converter for country code handling