2026
RAIS Publications
What does a research centre on artificial intelligence actually write about, and who writes with whom? RAIS Publications answers from the centre’s own record. It takes every publication that the Research Center for AI in Science and Society (RAIS) at the University of Bayreuth has registered in ERef Bayreuth, the university’s repository, and draws it four ways. I built it in October 2026.
Four views
- Semantic map: each publication placed by what its title, keywords and abstract say, so related work sits together across journals and disciplines, grouped into topics, with each publication’s nearest neighbours one click away
- Keyword network: the authors’ own keywords, linked when they appear on the same publication
- Co-author network: who publishes with whom, inside and beyond the University of Bayreuth
- Dashboard: publications by year, type, topic, keyword and author, with the full list
It currently covers 196 publications by 640 authors, from 2022 onwards.
Made to be embedded
Each view stands on its own and can be dropped into any web page: the centre’s site, a project page or a presentation. A “Copy embed code” button captures the current filters, so an embedded view can open on a single author, keyword, topic or period, in light or dark. The pages load no outside services and need no server or database.
How it’s built
A Python pipeline reads ERef’s export, merges variant spellings of keywords and author names, places each publication with Google’s open EmbeddingGemma 2 model and lays out the networks. The result is a static site that GitHub Actions rebuilds every week, processing only what is new. The charts use ECharts and the maps and networks MapLibre GL. The visualisations are adapted from the ones I built for AMIRA, here reading a simple data file instead of a database.