How CSranking Reveals the True Strength of Global Computer Science Departments

MIT professor Emery Berger recently launched the open‑source CSranking project, which evaluates the strength of computer science departments worldwide based on the number of papers published at top academic conferences.

The ranking method focuses on conference publications rather than citation counts, providing a clearer picture of a university's research capabilities and faculty quality; unlike traditional rankings such as US News, it compares institutions solely on top‑conference output.

CSranking is open source, inviting users to improve it via GitHub, and the author plans to incorporate citation data in future versions.

Each paper is counted once, with scores adjusted by averaging contributions among co‑authors and using a geometric mean across fields so that large and small research areas have equal weight.

Data are sourced from DBLP, and the website presents rankings in six tiers: top 10, top 25, top 50, top 75, top 100, and all universities. Users can filter by region (US, Canada, North America, Asia, Australia, Europe, Global).

The evaluated areas are divided into four blocks:

AI: artificial intelligence, computer vision, machine learning & data mining, natural language processing, web information retrieval.

Systems: computer architecture, networking, security, databases, automated design, embedded real‑time systems, high‑performance computing, mobile computing, measurement & performance analysis, operating systems, programming languages, software engineering.

Theory: algorithms & complexity, cryptography, logic & verification.

Interdisciplinary: computational biology & bioinformatics, computer graphics, computational economics, human‑computer interaction, robotics, visualization.

Top global rankings are dominated by US schools: Carnegie Mellon University, MIT, and Stanford University lead, with the top ten all from the United States. The 11th position is held by Singapore National University as the leading Asian institution.

Among Chinese universities, Shanghai Jiao Tong University ranks 27th with a score of 4.6, while Tsinghua University shares the 31st spot with the University of Edinburgh and UCL.

The platform allows arbitrary combinations of regions and research areas. For example, selecting “Asia top 10 + AI” yields Tsinghua University first (average score 23.9), Hong Kong University of Science and Technology second (17.8), and Singapore National University third (16.8).

Within Tsinghua’s AI field, the top three authors by paper count are Dr. Zhu Jun, Dr. Sun Maosong, and Dr. Tang Jie.

Changing the filter to include all computer‑science subfields shifts the ranking: Singapore National University leads with an average score of 6.3, followed by Shanghai Jiao Tong University and Tsinghua University.

This flexible tool makes it easy to explore the number of top‑conference papers and author contributions across institutions, offering a useful, though not definitive, indicator of research strength in computer science.