Data Science for OSPO: OSS Trends Found from Developer Activities
Talkat OSPOCon @ Open Source Summit Japan 2023 · 2023-12-05..12-06· at 16:05· Reception Room 2 (3F Level), Ariake Central Tower Hall + Conference, 3-7-18, Ariake, Koto-ku·
summary
Uses commit data from active OSS repos between 2020 and 2022 and clusters co‑occurring keywords with Infomap (compared to a Ruben algorithm), producing ~200 clusters: top ones are JavaScript, machine learning, cloud/Kubernetes; growing clusters include Java/microservices and databases, while C/C++ and game clusters decline; small but notable clusters show COVID‑19, quantum computing, and bioinformatics/genomics. Correlates repo commit timelines with developer share rates using a 0.4 threshold to detect substitute vs complementary projects and maps GitHub follower networks with degree and betweenness centrality to find key developers and sponsorship trends.
AI-written summary · about
chapters (7)
- 0:00Data science for OSPO
- 1:56Evidence-based tech policy
- 5:19Data science in OSS research
- 8:19OSS keyword clustering
- 13:33OSS developer shift analysis
- 17:44Key OSS developers
- 23:16Summary of analysis
related talks
39mA Data Analysis Approach for Contribution-Centric Open Source Strategy Planning
01h 04mWhat can we learn from 750 billion GitHub events and 42 TB of code
30mOSPO Data Pyramid
38mQuantitative Method for Open Source Contribution Value and Its Impa...
10mO’Reilly Radar: Open Source Technology Trends
19mTop datasets for top DevRel results
15mWhat's Next
57mUsing developer-centric data to predict, prioritize, and improve App Security Outcomes