The underlying research examines the socio-technical and regulatory landscape of cancer and health research from a Science and Technology Studies perspective, using Actor-Network Theory to trace how human and non-human actors co-produce scientific knowledge — and how AI stops being a mere tool and becomes an actor in its own right.
Two networks were extracted from the repurposed PubMed datasets (DB1 and DB2, cleaned in OpenRefine) with Table 2 Net, connecting two components: keywords and authors. The resulting graph was spatialised in Gephi with ForceAtlas2, a force-directed layout that treats nodes as charged particles that repel one another and edges as springs that pull them together, until the system settles into an equilibrium that can be read.
Filters then set a meaningful degree range, and node and label size were ranked by occurrence count: the more often a keyword appears, the larger it sits on the map. Position is meaning — a tight group of nodes is a cluster, densely interconnected internally and sparsely connected to the rest of the graph.
This interactive version reads the node positions, sizes and colours directly from the final Gephi export, so the spatial arrangement you navigate is the one the algorithm produced. The forty labelled keywords are the ones Gephi printed; the smaller unlabelled nodes are the long tail of terms and authors below the label threshold, and author names were not carried in the exported image.
Sources: Gephi (open-source community) · Table 2 Net, Mathieu Jacomy & Sciences Po médialab · ForceAtlas2, Mathieu Jacomy & Tommaso Venturini