Research discovery · GitScholar
The AI Almanac: Discovering emerging AI research through papers and code.
- Emilien Guandalino1
- Lorenz K. Müller1
- Beatrice Alessandra Motetti1,2
- Konstantin Berestizshevsky1
- Lukas Cavigelli1
Huawei Computing Systems Lab (CSL), SwitzerlandThe AI Almanac is an online research discovery platform. Explore recent papers, follow emerging ideas, and see how research topics evolve, using signals from academic publications and GitHub activity.
Scroll to explore the interactive story
Research leaves a trail.
A new paper is only the beginning. Its authors, citations and code create connections that help put the work in context. Build the picture in three steps: start with papers, add authors, then connect the code.
Start with the literature.
A paper cites another paper. These links describe the academic conversation, but citations take time to accumulate.
Conceptual illustration of the dataset structure.Dataset structure: GitScholar, Section 2.
Three emerging papers.
A fresh look at the top three papers in Almanac’s seven-day view. The latest available ranking is requested automatically each time you open this article.
Latest available 7-day ranking. Retrieved on opening; Almanac releases data weekly.
Finding your next read…
Follow an idea
as it takes shape.
The Almanac groups ideas into research families and tracks their evolution. Its weekly snapshots offer another way to explore a field, alongside individual papers.
Meet GitScholar.
Can activity around code help identify influential research earlier? GitScholar studies this question by connecting academic records with timestamped GitHub engagement.
Rounded dataset counts reported in the paper, for its June 2026 snapshot.
Predict early. Evaluate later.
GitScholar tests whether early GitHub activity helps predict citation attention one year later. Timestamped records keep later events out of the evidence available to the model.
Reading the signal.
A reason to take a closer look.
The model estimates stronger future citation attention from the signals available. Open the paper to understand the contribution: the score does not establish correctness or scientific quality.
Read the paper.
Explore the data.
GitScholar: A Dataset for Predicting AI Research Impact from GitHub Engagement
Sources: Almanac’s methodology, the GitScholar paper and dataset documentation.
Back to overview ↑