Bibliometric project

Before you run the project, please download the data folder, unzip it, and add it to the project root folder.

pip install -r requirements.txt

---

Section A. Bibliometric insights from CREDS research outputs

• A1. Basic data cleansing, author id and name disambiguation
  • remove noise
  • merge multiple ids
  • unify names
• A2. CREDS members author-topic Sankey diagram
  • visualise author-concept distributions
  
• A3. CREDS members author-topic & correlation heatmaps
  • find who and who share similar research interests
  
• A4. TSNE visualisation for CREDS members
  • show every CREDS member's position on a 2D plot (research interest based)
  
• A5. CREDS research concept vector
  • A pooling vector representation of CREDS research direction
• A6. Retrieve citation data for CREDS members
  • Data collection via API
  • for workflow purposes
  • generate 'data/ref_author.RDS' and 'data/ref_concept.RDS'
• A7. How do CREDS members cite each other?
  • An internal citing map of CREDS members
• A8. Who do CREDS members commonly cite the most?
  • The top commonly cited authors of CREDS members (common_cited_authors.xlsx)
• A9. What concepts are cited most by CREDS members?
  • The top commonly cited concepts (topics) of CREDS members (common_cited_topics.xlsx)

Section B. Construct Australian benchmarks for Education research

• B1. Citation performance of CREDS members
  • CREDS members' citation performance
  • total citation per paper (pp.), yearly citation pp., and 3-year citation pp.
  • Box plot of the three indicators
• B2. Constructing a benchmark based on the concept Education
  • Education topic, and have AU authors.
  • Scalability issue here, the API only supports 10,000+ records per request.
  • We can use the year-separate approach for records < 10k per year
  • Or we can try to download the database snapshot at a bigger cost (https://docs.openalex.org/download-snapshot).
  • Perform analysis on the three citation indicators and compare.
• B3. Constructing a benchmark based on the related works
  • Use the related works as a benchmark.
  • Perform analysis on the three citation indicators and compare.

Section C. Zotero dataset analysis

• Get the data (article DOIs) via Zotero API (167 fetched)
• Citation map visualisation (154 connected papers)
• major path analysis (three important citing paths)