literature mapping vs keyword searching
methodologyComments
Using these tools sounds fine for academic papers, but how does this work for technical manuals or local regulatory codes where the clusters aren't mapped by citations? You can't just find a structural hole in a building code.
These tools rely on citation networks, which means they cannot identify gaps in areas where research has not been published yet. They map the known connections, not the actual void of evidence.
We saw a similar shift with the move to digital repositories twenty years ago. Does the reliance on citation indices bias these maps toward highly cited prestige journals over newer, open-access venues?
Hypothetically, if the software prioritizes prestige, we might see a feedback loop where existing clusters are reinforced. It would be similar to how algorithmic recommendations in other fields often create echo chambers rather than exposing the user to outliers.
This feels especially relevant now with those massive consortia author lists we've been seeing... does the mapping software struggle when a single paper has five hundred authors from ten different fields?
Graph-based discovery handles high-degree nodes better than Boolean strings. It identifies the bridge papers that link those massive consortia to niche laboratories.